Many don't realize that there's an inherent delay in MIDI 1.0 to hardware. A single note-on messages takes apx 1ms to send over the wire, assuming it's the only thing being transmitted. If you use MIDI like we did in the 80s, using MIDI Thru on devices, you can get even slightly more variation. But more importantly, when you have multiple channels of MIDI data (multiple synths) then you end up with even more potential delay between note messages (up to 15ms more delay for a note on if you manage to max everything out). So even without the delays introduced by underpowered CPUs of the day, with MIDI, you already have some variable trigger delay in there. Of course, this only applies to MIDI, so post-1983 stuff.
Yes absolutely. What was especially strange about these tests though was that the random drift continued even when I took midi out of the equation and just pressed play on the drum machine. Also, I started to guess that the way the drum machines were implemented internally was actually serial (like midi) because nothing *ever* lined up and was always at least 1ms out.
I was wondering what caused the timing issue, I was thinking along the lines of US instruments being converted for UK 50Hz electricity. My set up must have some serious timing discrepancies as my MIDI set ups are a rats nest of cables. I wonder why MIDI never evolved into an optical system? probably cost!
@@AndrewAHayes Yeah. The original MIDI standard was about using inexpensive components that were super easy to integrate in 1983 because the CPUs all had serial integrated. Optical is more expensive and requires many more components. MIDI 2.0 is faster, and with the microcontrollers available today and used in synths, network MIDI 2.0 (spec now complete and being voted on) has a ton of promise for being the inexpensive super fast solution (also without the distance limits of USB)
@@Psychlist1972 Still no chords though, right? And you still get ‘gapping’ with the cc data because it’s relegated below the notes. People always say that they didn’t expect to be listening to suchandsuch a song forty years later. Well I certainly didn’t expect to be peering over tiny midi monitoring screens forty years later, trying to figure out if a channel is properly taking priority 😂😂😂
@@DistortThePreamp the MIDI protocol itself doesn't have any prioritization other than system real-time messages like clock. Everything else transmits as the sender wishes it to. Chords in MIDI 1.0 are serial, so if you somehow managed to hit 5 notes at the exact same time, there'd still be apx 5ms between the first and last, assuming no other traffic. In MIDI 2.0, they are still serial messages, but the timing is in the microseconds, not milliseconds so it becomes far less of a consideration.
An example would have been nice. You stated you tried it by recording it "normally," and then quantized(?) it, and there was a noticeable difference. Well, why didn't you share that with us?
Yes, without A/B testing then it's just a hypothesis. It's like "trust me, this is why". All music up until the 2000s was using crappy MIDI drifting timing, so I don't buy that is what defined the 80s sound, at all.
Yes. I’m really sorry about that. I made a previous version of the video which contained the full experiment but I couldn’t get it below 35 minutes and so, perhaps wrongly, I decided to remake the whole thing in a simpler way. The original experiments were done a couple of years ago, but the new experiments I did and filmed for the video worked as follows: - I sampled some of my drum machine sounds into the DAW - I programmed an identical pattern in (a) a midi clip in the DAW, and (b) my sequencer. (Note: I should also have done it in the drum machine which I have previously tested). - I recorded both. - Switching between the two records they sound the same, but they’re not the same as I then prove: - I take the kick-snare audio sum from both the midi audio and sequencer audio - For each of those I duplicate, layer with an offset of two beats so every kick-snare is under the previous kick-snare etc - I then flip the phase of the duplicate and show how with the midi audio it produces silence (a successful null test) and with the sequencer audio you get a phasey hollow sound that is noticeably different with every hit. - This proves that they’re different, and also different every time. And is also a perfect explanation for why DAW beats sound different to beats that were made in an old fashioned way using drum machines or sequencers. - I also reference some science which explained how, although humans typically perceive sounds that occur within 5ms of each other as occurring at the same time, our brains are highly attuned, albeit subconsciously, to the sound of phase shift because it’s how we perceive front to back, and from an evolutionary perspective it was an important part of staying alive when there were predators behind you. - Thus, in a slightly humorous way, one might say we have evolved to be able to distinguish between DAW beats and drum machine/sequenced beats, even if our modern brains can’t explain what exactly we’re hearing that’s different. - Finally I note that for people with proper training, they can pick out the differences and demonstrate with a blind A/B test. This last sequence was planned and demoed but not filmed. I’ve done a lot of thinking about this, and I am coming to the conclusion that I may entirely re-make this video in a few months with more of the science and demonstration. In the meantime, sorry for the hand waving.
This was a suggested video for me, and I’m quite interested. I appreciate the thoughtful response, so I’ve gone ahead and subscribed so I can see the follow-up. Greatly anticipated.
because he's just repeating the same thing all the yootoob experts have been saying five billion times...if you showed and a b the video would lose its impact
This video surprised me, namely that apparently so many people are unaware of midi humanizers. The M4L humanizer I've been using for about 10 years which was based on statistical findings of real human drumming published in studies like "Synchronization in human musical rhythms and mutually interacting complex systems".
Well this is true. I alluded to humanisers, right? (Or did I cut that bit?) I can’t remember whether a M4L humaniser can advance as well as delay midi, but if it can’t than the obvious workaround is to use track delay to advance the track and then double the amount of humanisation. Although as I’m writing this, it probably doesn’t matter that much - the main point IMO is capturing not so much the timing drift (which is what humanisers were intended to do) but rather the way the phase relationship changes between two overlayed sounds as a result of a differing drift in thei alignment. Great point.
So the timing feeling humanly natural doesn't matter as much as each drum hit having some unique quality caused by phase "issues". This is, i imagine, why people still refer to this era of music as rigid.
EZDrummer has this built in too. the funny thing is they also use real drummers and people sometimes it humanize that and it stuffs the sound up as it kind of doubles the effect makes it sound unreal
Ive always imagined this was the case - I first noticed it with the Linn Drum where there is a noticeable lag if you put two sounds on the same subdivision - only one sounds in time, the other one comes in late. If you listen to those early Prince albums using the Linn Drum the Kick is almost always fractionally late on the one. He then took that on as a key part of his sound with his live bands always dragging the one - its part of what makes his feel so funky. When I first got into sequencing I used a Roland MC500 which seemed to have a great feel. When I switched over to using Vision on an early Mac laptop it still seemed to but when I switched to a modern DAW and had sounds actually IN the computer rather than on an external machine, I found that if I quantised it was too mechanical - I had to deliberately leave in timing errors, especially on drum programming, to get a natural feel. It's brilliant that you've taken the time to actually create a comparison and get the evidence! It would be cool to hear its
Exactly. There’s another reason that’s peculiar to the Linn which is this: all of the samples had ‘dead air’ in front of them. There are lots of theories as to why this would be the case, but my guess is that Roger Linn wanted to make sure that the VCA would have time to open. However, what’s interesting is that as you define the samples the dead air (of course) gets longer. This means that the groove slightly changes depending on the drum tuning. And Prince often used very detuned claps and rimshots. I am fascinated to see/hear if the Behringer clone replicates the dead air. If it doesn’t, it won’t capture the Linn groove.
@@DistortThePreamp I didn’t know about the dead air, but that makes a lot of sense as to why those detuned sounds feel so late! I wonder if he was de-tuning primarily to get the timing effect and accepting the pitch shift as the side effect!!
The phase argument also makes me think about the oscillators in analog synths that just keep running, instead of restarting with every 'note on' like in many (but not all) digital synths.
A lot of the timing in discrepancies were also due to the fact that they recorded onto 2 inch tape and a lot of cut and splice took place inherently changing timings of different tracks Which, when it comes to tracks from prints, and those classic 80s anthems clearly added to the complexity of the finish product But as a user of hardware, I can agree that when you look at it, after recording, you can clearly see it’s about the groove I love your analysis ❤
Yeah, I was so surprised about this. I had assumed, like you're alluding to, that the timing discrepancies were to do with either tape stretching, or general clock drift. I was astonished to find that the drum instruments were drifting independently even when you just hit play, in other words with no midi involved at all. So when you add in SMPTE clock, etc etc etc, it all sounds very non-robotic.
Makes sense. The spec. for the MIDI 1.0 protocol guarantees a maximum latency - “response time” - of 8.3ms (at 30 frames per second). That's well within what would be comfortable for live performance, but it's not nothing. And it fits rather neatly with your observations of around 7ms of drift. It never occurred to me how it would affect the sound. But of course it will.
Oh I didn't know that! And of course 8.3ms = 16.6 maximum delta. I was aware that every midi note had to be at least 1ms apart, and there is no such thing as a midi chord (because it's a serial protocol), but I should obviously have read the spec. Duh.
Interestingly the Juno 106's voice CPU has a "cycle time" in which it updates all six voices and the "global" parameters (noise, master VCA, PWM, resonance, and the various switches) is 4.3ms, but by the time the assigner CPU has done its work and it's told the voice board over its MIDI-like internal link, it averages out at about 6ms latency between "Note On" message and sound coming out.
I'm still curious about recommendations for a modern hardware sequencer, however... this morning I wrote a script for Reaper that will randomize the placement of all selected items by up to a certain number of milliseconds, plus or minus. I use Reaper for my work so it's the DAW I generally noodle around in with music. I created a drum and kick track using an Oberheim DMX plugin, printed them to audio, which resulted in them all being perfectly on the grid. I then used strip silence to isolate all of the hits, then ran my script and entered 7ms as the value. It randomly pushed and pulled all the individual drum hits by up to 7 ms forward or backward. It definitely changed the feel of things. I think purely random may be the wrong approach, as it sounded a little chaotic. Maybe the better way is to, for example, pull all the kicks forward by 3ms, then have it push them forward or backward by up to 1ms more after that. To make it a little less random. I'm definitely going to play with it though.
So you sound like you’re properly mathsy, and therefore I don’t think it would be a random linear deviation around the mean. You might want to try a power distribution with an upper bound, but there’s another thing which is that none of the hits ever occur together and are always at least 1ms apart. That’s probably because the internal drum machine sequencer is modelled on midi and is serial. All lots of fun. Regarding hardware sequencers, I absolutely LOVE my Squarp Pyramid! Unfortunately they’ve retired it and are trying to move to their next gen model which looks great but is a bit physically big for me. So my recommendation is a second hand Pyramid. At some stage I’ll make videos showing my workflow, but suffice to say it works excellently for electronic pop and rock. Hit me up on my email in the description if you have any specific questions…
I’ve worked as a sound designer in the video game industry for almost 20 years. I take advantage of this phenomenon all the time when making transient sound effects. What I didn’t know was that this was happening all over 80s albums but it makes perfect sense.
Oh interesting! So you are taking identical samples and exploiting the fact that when they layer the sum sounds different because of phase differences? That's fascinating! And let me return the comment -- now I think about it, it makes perfect sense in your industry too...
@@DistortThePreampSometimes identical samples, sometimes similar samples. Sometimes I’ll cut off the first one once the second one kicks in, kind of like “choking” it in drum machine terms. We do this for explosives, gunshots, or anything big and transient all the time. Theres a huge variety of sound you can achieve when playing with very short amounts of time offsets. I remember when I first learned this, a more experienced sound designer taught me and it absolutely blew my mind how much a few ms could change the timbre of the sound.
Oh that sound designer would have been talking about exactly the same thing I was. It’s extraordinary once you witness it with your own ears. I’ve never done any game sound design but it sounds fascinating. Your description of ‘choke groups’ is obviously very applicable to music…
Fun. I drag my tracks in the daw a few ms to the front or back and then repeat till i get what i like most. But i should repeat it more later in the track 😎
Fantastic. Thank you. I was the Fairlight Series III CMI demonstrator at Peter Gabriel's 'Syco Systems'. The timing 'issues' took me to Australia and back, then we made different versions of the software for different clients, specifically due to the quirkiness of the Series IIx and Series III. The NED Synclavier was within about 1 ms, however many of my clients preferred the Fairlight 'timing', as Trevor mentions in his book. Here we are still working with the original Fairlights and I use them for specific jobs for making soundtracks / music productions - still! :-) Your 'Temporal Phase Shift' is great. Another important element is that the Fairlight had separate individual outputs 8 and 16. Therefore the calibration of each would be slightly different. So each individual note would be both temporarily 'phase' shifted, as well as the frequency / EQ curve slightly different on EVERY note, depending on how many output (or Nphony) was assigned to each multi-sample. If that makes sense?! In other words, over 50% of every hit record from that time, needed the Fairlight somewhere in the production to make it sound like the 80s. Emu and Linn also had their own quirks, as well as the Ensoniq Mirage. I hope this helps? Cheers! PS With the Fairlight 'Page R' or 'RS Page' timing priority was given to the top 'track' and became more 'wobbly' the further down the sequencer page. There are stories that I could tell..........
Roger - such an interesting comment! I had no idea about the Page R priority! That’s hilarious! What was surprising to me was that I could hear the alignment shift in the sound combinations well before the timing errors were apparent. I was a bit coy in the video (although I dropped a very large hint) but that was an actual multitrack (that I shouldn’t have) that I was showing of a famous number one produced by Julian (whom I’m scheduled to interview in a few weeks btw). You can literally see the drift. Some of the snares are ahead, some behind, etc etc. I think everyone has historically focussed on timing discrepancies as a rhythm issue, but of course it’s just as profound if not more profound as an alignment issue between sounds that makes every combination sound different. I’ve just been doing some snare sound design where I’m layering two identical samples and delaying one of them. As you would expect a delay of only 1ms produces a very noticeable comb filter. The Linn timing issues are fascinating. To me the biggest ‘secret’ about the Linn is that there was dead air at the beginning of the samples which of course got longer when the samples were detuned. So when Prince was detuning his rimshots he was actually changing the groove. And that’s before one analyses how accurate its internal sequencing clock was. If you feel like it, get in touch over email (address in the description). I have too many questions for this short reply!
@@DistortThePreamp Great. I'm in touch still with most of my old clients as there's 3-part documentary in production at the moment. I'm happy to chat via email :-) I 'curate' a number of old machines...
I've been recording an 80's disco/goth fusion EP, and i made this discovery myself recently. I'm sequencing on a 606 as i love the workflow. Recording a bar at a time, and after I'm done I'll adjust the daw's grid to match what I've recorded. The looseness of the timing is an interesting thing. Apparently the 606 actually has capacitors in circuit specifically for this, not randomness, but looseness. This video really did put my mind at ease, i was working around all these inconsistencies, my ears were telling me it's fine but my eyes were not. You know what they didn't have back in the 80s? Graphical waveform readouts.
Exactly - no graphical waveform readouts 😂 That’s why I wonder whether I’m one of the few people to ever have done these tests. But I’ve decided that I very much like the sound of the drum machines.
Reason studios has a drum module called “Umpf Retro Beats” that has vintage 80’s drums and the gated reverb. It also has a “regroove mixer” that you can put your drums on different grooves and you can “slide” the midi so you can get this drifting effect.
Congratulations for this video. It really is helpful and we all can see you have a beautiful passion when speaking about the 80s music and overall sound. Thanks.
Loved your obscure PSB's reference and for letting me know the song. This and the NO code made me subribe immediately. As a producer, this stuff is as addictive as crack.
I’ve been using an Akai MPC60 mk1 as my main midi sequencer for about 15 years now with a studio full of synths, samplers and midi kit. It’s tighter than any computer based midi sequencer. Even better than my old Atari St from 30 years ago. The Atari was tighter than any Mac or PC based sequencer but the MPC has the edge. You can offset the timing on the MPC too. It gives such an organic feel. I’ll then multitrack the audio into the computer. Slaving the MPC (can I say that word now?!) to the computer via midi clock.
I do a similar thing - I start by programming in my sequencers, then I track in and slave back to the sequencers using compensated beat clock. It’s all right enough. But nowhere near as tight as putting stuff on the grid inside a daw and hitting the render button. And this IMO is a good thing.
I was stunned when I discovered this. As I said in the video, I spent about a month trying to solve the ‘problem’ till I did, and then realised it wasn’t actually a problem. I honestly don’t think even the 80s producers know this. I was out with someone who used to work at Trevor Horn’s studio with all of the usual suspects, and it was certainly news to him.
I just noticed this yesterday myself when I set my DAW to 140 BPM, the drum machine synced up nicely, but sometimes it dropped to 139 BPM for a fraction of a second then jump back to 140. It happens relatively often.
Creating 80's megamixes though is a nightmare for that very same reason.. I have to warp the tracks in most cases every couple of bars and then offset afterwards once I have introduced newer drums to bring back that live feel. It's funny when you think about it. When there were imperfections in beats, everyone would clamber for perfection. Now we have it? Look what we want to go back to. (I say we, because I am guilty of it too)
Oh thank you! Old fashioned vocal processing, particularly country, is something of a lost art. Technical limitations made it impossible to do the kind of processing and overdubs that became standard even by the 70s. So if you wanted to change the EQ, you’d better select a different mic. Similarly, rather than rely on compression, the recording engineer would learn the song and literally ride the vocal fader (that sets the recording level) during the singer’s performance. It was a completely different world.
Since 30 years ago not only we always loved the famous MPC feel, knowing it’s swing factors, and realizing the MIDI transmission imperfections like you very well mentioned, but I also noticed the various synths had different speeds to receive busy MIDI info and to generate the corresponding sounds. And it doesn’t stop there… the different sequencing softwares had different timing characteristics! I remember loving the feel of the Opcode’s Vision and I used it on many many records, until it was sold to Gibson who sadly killed it. I never really understood why they bought it! Anyways, after making records all my life, I rarely get impressed with what I see people saying, but you truly are a genius. Your videos are very unique, intelligent and inspiring!
Thank you! The MPC is really quite special sounding. Really great workflow too that doesn’t get in the way of creativity. I don’t know much about the new ones - it’s potentially a bit over-complicated - but there are still a lot of great records being made on the 2000XL.
I remember recording on my computer back in the day using an E-mu Emax II and an Alesis MMT-8, and I never quantized my music. When I recorded my first album in the 2000s, the trend was completely different-it was all about achieving a perfect sound. Projects like Thievery Corporation, Jazzanova, and those on the Naked Music label were focused on sounding flawless, so everything was quantized. I decided to mix both natural instruments and quantized elements to give the music a more 'alive' feel.
Have a look at the “Human” vst by Mixed in Key . I enjoy setting instrument part variations in my DAW (+ or -) 5-7ms per drum instrument part and then add Human to get the variation in timing. Saves a ton of headroom and adds that secret sauce to keep your ears interested.
DAMN your channel is badass! I saw the title and thought "I imagine it's gated reverb? I'll be back to watch this soon" but this is SO much more than I could have imagined 🔥🙌
Oh fantastic! It took me a while to get the title of this right. And you wouldn’t believe the pushback I’ve had - it’s easily my most co traversal video (and I really never try to be controversial as I think it’s bad entertainment). So I’m SO GLAD you liked it :)
Back in the early 90s when i was studying audio engineering one of my teachers told me: record your drums by hand and loop or copy paste each 4 measures, but do not quantize. Or you can quantize the kick but do not qualitize anything else... duplicate and loop but do not quantize, you need those harmonics around so it can sound organic.
What a great video. You have unique talent in explaining thngs. I always assumed that the majority of 80's drum machine parts were either programmed in the machine, or alternatively quantized after being played manually, but I later found out that artists like Michael Jackson's musicians had the Linn Drum parts played by hand and not quantized afterwards. Other bands played live drums on top of a programmed drum machine (Fade to Grey etc). I would have thought that drum machines have their own rigid clock like on a DAW sequencer, but not so! There is a temporal phase drift as you correctly mention!
Thank you! I was as surprised as you that the clocks weren’t and aren’t rigid. From what I can gather it’s simply a function of an underspec’d processor because the underlying quartz timing is very accurate. So the question becomes why on earth would they underspec the clock? And I’m guessing the answer is that they felt the timing was ‘good enough’ and didn’t want to spend the extra money. I’m not complaining, but it’s pretty hilarious. And it’s even funnier that my modern Squarp Pyramid sequencer keeps time like a drunk Uncle at a wedding.
Would have been so great to hear a comparison of the “hardware” sound vs computer sound so I could actually know what the special timing drift actually sounded like. I thought for sure that since you had the isolated tracks and could probably replicate the same beat on a computer, that at the end you would have played it back to back. Interesting video for sure!
Yeah, I’m really sorry about this. Loads of people have made this point. It was my mistake really - I had done a version of the video with exactly that, but it was over 30 minutes and I was worried that it was too boring. I will re-make this video in a few months in a different way and show the full experiment :)
Dear Sir, Please accept my apologies for my previous comments. I have watched your video again and realise my late night brain was misunderstanding the points you were making. You have explained and illustrated very eloquently and succinctly a huge amount of technical and creative information that helps us to understand what makes 80's music so great. As a content producer myself (not on this platform, and not music) I enjoy the indulgence of a little imperfection here and there. I was a young teenager in the 80's, and often felt that some of the pop songs were a bit "thin", but when I listen to them now (in my 50's), I realise that they have a complexity and sophistication way beyond what I was able to appreciate at the time. I did enjoy much of the music that was generated in that era, including many of the bands that you reference in your videos. I appreciate your wisdom, and have accordingly subscribed and enjoyed a number of your other films. please continue to share your insights with the world. Kind regards. PS, as a young teenager I was involved in a community "radio station", which actually involved cutting and splicing tape. later, in video editing, I came to understand that you could take advantage of the natural softening of a (electronic, not physical) "cut" on magnetic tape to make an audio cut more natural. I think that relates more to part 2 of your film. Thanks again.
Dear Tony, thank you for these very kind words! Your experience of cutting and splicing tape is interesting. Were you doing something like deck-to-deck editing on U-matics, or Betacam, or something like that?
@@DistortThePreamp The cutting and splicing of tape refers to when I was involved with a school "radio" station when I was around 13 yr old. We did a 20-30 minute programme broadcast over the school tannoy system on a Friday lunchtime. Music requests, shouts, news etc. So it was 1/4 inch audio tape. we would use iron filings on the tape to get the exact cut point on the signal path. Razor blade and splice block for the (diagonal, for the cross fade) cut, sticky tape to join the bits back together. long time ago, so I don't really remember a lot of the details. We'd probably call it a podcast these days. The school also had a Black and white 3 camera TV studio that I was involved in a couple of projects. Probably what sparked my interest in video, though it would be around 20 years before I got to that in a professional sense. It was all analogue at the time. I remember one of the techs that helped run the studio showed me that if you set the vision mixing desk up in a certain way, you could use the cross-fade control as a "threshold" for a cut, and if you got the sweet spot, it would split the screen vertically with a random zig-zag pattern. That blew my mind. And was very much "a la mode" The video thing I mentioned was Betamax SP. Same tapes that run on Betamax domestic players, but at twice the speed for broadcast. not really possible to physically "cut" that in the same way as audio, as (I'm sure you know) the video track ran on a slightly curved diagonal across most of the track, while the audio tracks ran in parallel along the side. So, to come to your main question, in my early days as an assistant editor (video), I was trying to do a really tight audio edit on a Betamax SP tape by setting up the audio mixer to do a one frame (25fps) cross fade, matching the vision cut, triggered by the edit controller. But the audio mixer couldn't do a cross fade in less than 3 frames. The supervising editor said "just make a cut ! ". with some disbelief, I tried it and it came out as a perfect soft mix. After some reflection (and I still marvel about this to this day) I realise that the malleability of the physical structure of the materials embedded in the tape would only partially move, creating the 1-frame dissolve at an atomic level. Crazy nuts stuff when I think back on it ! PS, I hope you don't mind, I have deleted my negative comment from a few days ago. Kind regards. T
I can’t replicate the results. For my first experiment I used Reaper's “Dynamic Split Item” feature to split up the transients by threshold on a recording straight out of my RX5. It was accurate to within 50 samples. Not satisfied, I tried manually, with a similar result. Then I thought about how the 606 has a reputation for inaccuracy so I recorded some of my Behringer RD-6, amazingly it was even more precise. Here are the results in milliseconds I got by manually slicing at the sample level. For the RX5: 0:00.249 0:00.248 0:00.249 0:00.249 0:00.248 For the RD-6: 0:00.344 0:00.344 0:00.344 0:00.344 0:00.344 That’s right, the inaccuracies can’t even be measured in milliseconds with me doing manual slicing. Am I misreading the Reaper figures, are they not milliseconds? I’m so confused.
Haha - thank you! I was so staggered that it even existed. I mean they’ve all got quartz clocks. But turns out the clocks for each instrument drift independently anyway. And the irony is, it sounds fantastic.
Am about 75% through, you've still not revealed your insight (predict the internal clock of the sequences to blame), and you've presented a solution to an issue I was having to remove the machine like exactness of the modern computer. Slower clock speeds. And here was me trying to programmatically add a random drift of 1% to my tempo. Thank you 😊 🎉
Oh fantastic! I still don’t really know why all the internal clocks are so bad - not that I mind, because I far prefer the sound. But my quest continues…
Great video, and I think you're right!. I'm curious about your recording workflow... assuming you record into a DAW, do you sequence everything using your external Pyramid sequencer and then record it all in one go? Or do you sync the sequencer to the DAW which would then allow you to do as many takes as you want? I have a Roland MSQ 700 which I can sync to my DAW for start/stop. This would allow to me program a sequence in the MSQ 700, sync the MSQ 700 to the DAW, record the sequence, program a new sequence, then record it, and have it all line up in the DAW. Is that what you recommend or would syncing to the DAW "fix" the "problem" and ruin the effect?
That’s pretty much it. I’m going to make a full video at some stage, but basically I start by tracking in the programmed parts from my pyramid into Pro Tools, then I do a tempo map in Pro Tools based on the kick drum, then I switch the master clock to Pro Tools so now the DAW is driving the clock. This means that if I need to do a sequenced overdub I can and everything lines up. I’m very familiar with this workflow in Pro Tools but I never really got it to work properly in Ableton. I am guessing that it would be possible in Logic and the other main DAWs…
The physical distance that MIDI data AND audio signal travels over the various wires between all hardware devices will also create its own natural lag, however minuscule, providing a nice sense of "timing" fluctuation as well. All clocks are a little different. I used to perform my live sets (DAWless) back around 1997-2003, like a DJ would spin records. No midi connecting any devices, only two Roland Synths (XP-80 and MC-505), and my natural sense of timing for triggering the starts of my next songs. It had the benefit of sounding amazing, or going horribly wrong, but it was human!!! It makes all the difference. 🙂
Yes. Audio cable lengths are not all that significant becasuse they’re traveling at the speed of light, but audio through the air is obviously about 1ms per foot, and midi messages are spaced at least 1ms apart and, given that midi can’t accommodate chords) this means that a four note chord contains a spread of at least 3ms. As you say, this can add up to quite significant alignment issues that simply don’t exist if you programme inside a DAW. It just sounds different.
I’ve noticed this on extended 12 inch mixes where there’s lots of muted tracks and solos, you can hear subtle “mistakes” that make it feel more human even though it’s electronic, I used to put it down to processing power being a bit slow with tiny lags
You’re absolutely right about 12” mixes from the 80s. Not to descend into a history lesson, but the commercial reason they were made is that the chart rules recognised multiple sales (up to a limit) as separate units sold. So if you could persuade a punter to buy the 7” and 12” then you would notch up two sales. Therefore there was huge pressure for 12” versions to be produced of everything and very little budget to do them properly. I know several producers who used to make them in an afternoon just by basic razor blade splicing. They would mute the vocal and let the mix play, then add a middle bit, then repeat from the beginning with the vocal added. Or something easy like that. I know Tape Ops who were told to “do the edits” because the producer couldn’t be bothered, and sometimes the edits weren’t fantastic. “Shake The Disease” (Depeche Mode) has a very bad edit in the middle of the 12” version that I always found annoying.
Trivial to add a bit of randomness in DAW drum tracks - generally called 'Humanise' or something similar. And the delay between voices in a hardware synth is not random - any delay between the first audio 'attack' of each different sample relative to a different one is constant for every beat. They do not drift around.
In Logic some native plugins like the Drummer, have a humanizer function. That helps and there is a free Plugin from Baby Audio, named pitch drift, that helps to get rid from the steril sound of samples.
Recently I have been listening to early Soft-Cell, Ministry and Depeche Mode, and was thinking how organic and lively most of that music sounds, and even thought it’s mostly drum machines, synths and sequencers, you can tell that it’s not hard quantised at all, to an extent that it almost feels like those instruments were played in real time by humans because some of those sounds really have a groove. Now I know why!
Some stuff might / will have been overdubbed, played by hands, onto the sequenced basics. Example: Basic Drum Pattern sequenced. -> Cymbals and tomtoms "overdubbed", by pushing the buttons by hand.
Great video, really interesting! Just to play devils advocate, the latest version of ableton live allows you to 'loosen' the quantized notes by a random amount within a specified range. Would that not have the same effect if you programmed your drums in a piano roll? In Cubase there is an option to similarly randomise velocity with a range which i use on things like pianos just to give them a bit more of an organic feel as i can't play well at all so usually rely on hard quantizing, then using this to add some humanity back in :)
Yes, it basically would! I was overstating for comic effect when I said the computers were dead - you can absolutely do what you’re suggesting, and in any case ‘grid music’ can sound fantastic! Having said that, I truly believe that this sound (ie the way the sound combinations change because of micro-alignment issues) is fundamental to why 80s music sounds the way it does. By the time the 90s came along it had been ‘fixed’. And the issue didn’t exist in the 70s.
Thank you! I am getting sooooo much grief, and rightly so, for not including examples. Ironically I filmed them but ditched them because I thought they were too boring. That’ll teach me.
I tried measuring my Yamaha RX5 in reaper, it's accurate to within 50 samples at 44.1 kHz. I also saw an interview with Roger Linn where he said the groove in his drum machines is a myth and they are dead accurate, although the MPC does have swing. From what I'm reading only 606, and 808 have any drift and obviously the older Roland boxes from the 70s. What I want to know is when people were using things like the BBC/UMI sequencer what was setting the clock? And further, how often did midi jitter come into play?
One would think that digital audio workstations would have an option/plug-in to introduce random tiny timing imperfections to the individual rhythm tracks in order to give it that feel.
For sure, especially with the Synthpop in the mid-80's, Freestyle and very early House and Hiphop, there was this sort of "drifting" that's very "80's". I didn't know it was early midi, I thought it was just tape or something to do with tracking but okay, that totally makes sense Thanks for the insight, it's definitely heard but awesome you brought it forward from the subconscious, thank you so much for the talk
very interesting video! i love vintage gear, but most DAW’s have a “humanize” feature that will anti-quantize midi notes by introducing a little random variance. no reason to reach for outboard sequencers unless that work flow is genuinely preferred or some other juice is being sought after
Ah, that’s a good theory, but actually that’s not it. That could in theory produce a drift, but not an alignment problem. In other words the instruments wouldn’t move *against* each other. Also tape running at 30ips (or even 15ips) doesn’t drift that much. Meanwhile, one of the old Fairlight team has actually confirmed that there were indeed timing errors! Really appreciate you watching though, and taking the time to comment :)
@@DistortThePreamp Thank you. I was speaking from experience. I tried to remaster something off 4 track cassette and recorded each track separately into my DAW and they don't line up. Also, sequencers back then did drift as well when using CV instead of Midi which was still brand new at the time.
@@spacecitysprockets ah - 4 track compact cassette certainly does drift. And tape stretch, as I’m sure you know, is something of an occupational hazard. I remember we were all advised to not use BASF because apparently they stretched more. However, those cassettes are only running at 1.875 inches per second which is 16x slower than an typical multitrack. It’s those that I’m saying don’t really drift, assuming it’s properly maintained and setup. Certainly the SMPTE codes were pretty stable at 30ips. Very interesting comment though.
@@DistortThePreamp Thanks for answering my comments! I do remember in video editing suites during the 80s, we could only afford 1 generation on 1" tapes, so it was very touch and go to do layering (overdubs).
Great identification of the situation and concepts to visualize what hardware timing gives us - but there are plenty of tools in a DAW to deal with this issue - moving events in any direction you want. Of courre that could be tedious, but there are simple ways to adding timing randomness... As someone who has been sequencing since the days of the Atari ST - In general with MIDI sequencing and computers, the PPQ resolution (or setting if available) also determines where events happen in terms of MIDI timing resolution.
This is why I always prefer to record all my parts with midi into Ableton or any daw and DIAL in the amount of quantization for each part. It is almost an art of its own but it is very possible and rewarding with modern daw. A percentage of x% difference quantization can make or break a part based off a really good human recorded part. And no recorded part is ever the same so the quantizing is always different.
Love it! I recently decided to become a music producer and a DJ and among the first thing I acquired besides Pioneer FLX10 controller was a Roland TR-8S. I knew I was going to use Ableton for production and live performances, but my gut feeling told me I needed hardware instruments. My latest purchase was a Casio Privia S3000 with 88 fully weighted keys. Words cannot express how much I love that thing! A full piano keyboard lets me explore the full spectrum of harmonies as I try to find happy accidents while jamming along to my looping track.
Oh brilliant! I think the TR-8S is a great choice. With drum machines I’m a huge fan of trying to ‘perform’ with them - using mutes, solos, filters, etc. And the TR-8S is really good at that. My top tip would probably be to go easy on the fx. They sound like ear candy in headphones but live I would use delay as an occasional effect and reverb pretty much never. Obviously the filter should be used all the time though ;) And the S3000 is great. I have a proper 88 key piano but it’s not rigged up in my studio because I don’t have space for it. And I really miss it. If you’re a ‘player’ you really need those extra octaves. Let me know how you get on!
@@DistortThePreamp you know what else TR-8S is great at? Being an Ableton controller. You basically get up to 11 channels with dedicated volume fader, filter knob, reverb and delay! But unfortunately I have mot found a way to launch abd stop clips with it, so I am getting a launchpad mini next.
@@riccello oh that’s interesting! I used to use the Akai APC40 before I went dawless. It was really fantastic because not only could you launch clips but it had tight integration with things like effects racks. So from memory you could pre-map effect chains to racks and macros and then control those very easily. The rotary controllers were also very well done - infinite turning with leds to show values etc. It also had the A/B function implemented which not many people use in Ableton but which I used to use quite a lot.
@@DistortThePreamp btw, I was so inspired by your video, that I researched a way to simulate random phase drift in Ableton. Basically, you can do two things. Fiest, you can Add a delay to each channel and set a range from 1 to 20ms. Then you can automate it using LFO. You can also randomize LFO rate. Second thing you can automate the same way is phase shift between left and right channel.
Ah - randomising the LFO is a great idea! 20ms sounds a bit much though. I would set the max to maybe 12ms. And maybe have it changing value every 16th note. And obviously apply it separately to every drum. That’s a really interesting solution. An additional tweak would be this: suppose you’ve set the max drift to 12ms, then you could also set a negative delay of the whole track to 6ms. So that the note was drifting around the grid line. And I don’t reckon you’re gonna have to do anything with the phase - although it would be interesting. But the mere fact that the instruments are changing their alignment slightly every time is gonna make the whole thing sound different to a DAW. As I think I said in the video, it wasn’t so much that I could hear the sound, it was much more that I could hear the contrasting sterility of the strict grid.
It's not too dissimilar to what the Beatles did with their ADT (artificial Double Tracking) where they'd take the same vocals and bounce it to another track, but with a tiny delay to thicken things. Sure, in those cases the delay is enough for you to notice but it's the same sort of principle. I'm not sure how or why The Pet Shop Boys did this in this case - it might be part of their MDI chain causing delays, or it might be how they edited takes together or something (I've also had hold of some of their multitracks in the past). Either way, serendipity and all that causes great things. There's always that saying of make space in your music to give it space to breathe. So it's not only the case with long periods where other instruments can come in, but also the more subconscious stuff like this where a tiny delay that we can't instantly perceive can change the sound signature enough, much in the same way as playing with a phaser is tiny in amounts but you can quickly hear the difference.
Hahaha! I see what you did there ;) I couldn’t possibly confirm or deny what multitrack I had up. But really there were and are soooo many things that cause drift in the old analog workflow. Even CV/gate which everybody seems to believe is “rock solid” isn’t rock solid at all - of course the triggers transmit at the speed of light, but there a small computer spitting out the trigger timings and, well, turns out it’s a bit wonky. You’re absolutely right about phasing. I still maintain that I can hear the difference in the way a snare sounds against a kick because of alignment changes waaaaay before I can hear the actual changes in timing.
I didn't want to say it, but glad you noticed :) Yup, I'm someone that for whatever reason dones't like using DAWs anyway. I trained at Salford Uni back in the early 1990s in audio engineering, so this was all original MIDI and Atari STs (if you were lucky) or analogue tape. I think I just prefer being more hands on. But yeah, there's always lag on such things, especially when you extend the chain more and more. I do find it laughable that people call CV/Gate rock solid. They clearly weren't alive back in the day and had to deal with timing issues :) By the way I love what you do - I appreciate talking about things that most people just well, don't.
Cool video! A few month ago I stumbled across the same problem and was astonished to see this. But I was editing the drums in a midi-track in the box (quantised it), played out to my RX11 and recorded it back via the seperate outs of the machine into the computer. Same effect. So my guess, it's somehow the midi-protocol that causes this. Or it is the "Midi-Synconisation-Protocol" that I've set to 50% default in the Computer . . . however, I know now about this issue and use it. Much better results. Thank you for your work. Greetings from Dresden.
@@le49exileaudioproduktion59 hey! So I thought it was the midi protocol too until I tried it with no midi at all. In other words just having my R8 spit out the drums when I pressed play. But whatever the reason, it sounds great so I’m not complaining!
Fascinating video. I think I have noticed these subtle changes in timing, but never put my finger on it. "Temporal phase drift", that's a very clever term. I follow a producer called Mirko Hirsch who does a fantastic job with some of his songs at remixing songs into 80's style.
Fascinating theory! Thank you for your great analysis! I think it makes total sense, and it's something I've considered also. what I don't really know yet, is a good way to get this done in my DAW. Any advice would highly appreciated!
Ah yes, there are lots of good ways. If your DAW has a bumanise function then that would work. For example, in Ableton you can use a Max4Live device called Humanize, stick it on every midi channel, and make sure it’s calibrated to move the midi notes within a +/- 7ms range (which might take some testing). Logic has a similar function built it. There are other methods, but I think those are the easiest! Because of my own personal workflow I’m a huge fan of actually rendering out the audio before mixing so there’s no longer any midi in the session, and in this case you would be rendering out humanised midi. This makes sense to me because it would mean you were taking a snapshot of the randomised timing drifts rather than having them be different with every playback. But that’s just me ;)
@@Tt-yr9ri Oh - absolutely let me know how you get on! Again, I strongly encourage you once you’ve found the right sound to ‘perform’ it in to Cubase as audio. In other words set the sequence running but if possible make some parameter tweaks while it’s playing and capture audio not just an automation line. Obviously preserve the midi, but then try to work with the audio. IMO committing in this way always helps a production ;)
@@DistortThePreamp It actually worked :D. There's something called Logical Editor i Cubase; it has presets to offset MIDI notes. Now it sounds a bit more as if my vst synths are performed by a band :D.
Oh brilliant! My top tip is not to overdo it. A very small amount goes a long way. I don’t know if you could measure the effect, but my recommendation is that every instrument, particularly drum instrument, moves a maximum of 7ms in each direction, and of course they have to move independently. That’s not to say that larger drifts won’t have an interesting effect too! But to just get the thing where every combination of kick and snare is technically different, you only need +/-7ms. Very exciting that you’re trying this though!
Ah yes. I had a wizard cast a spell which means that the mic no longer needs to be pointing at my mouth to pick up perfect audio. Very useful for UA-cam videos.
Great explanation! I don't have the technical knowledge to understand the time deviations in drum machines, but I know that nothing in nature is perfect. Our primitive instinct picks up on these imperfections, which gives us a sense of "trust," while things that are too perfect can raise suspicion and cause rejection. That’s why those subtle imperfections are more pleasant to us; it’s a clear example of the uncanny valley effect. Cheers!
That’s exactly right. My theory is that even though we can’t hear the time differences, we can sense the sound differences because we have evolved to sense minuscule sonic changes arising from alignment since that’s how we detected sounds coming from behind us (and avoided getting eaten). FWIW all the producers I showed this too (some of whom have won Grammys and been making records for 40 years) were incredibly sceptical until they tried it in their own studios with their own drum machines.
I've never liked the sound of drums being exactly on the drum grid, so I have always played my drums by hand and then used midi quantize around 60-80% (until it feels right). I'd imagine the resulting time variations are roughly similar the 3-7 ms variations produced by hardware sequencers, albeit perhaps less random.
Yes I think that’s right. Even a really tight human drummer is a few ms out. I mean, let’s start with the fact that the snare is closer to their ears than the kick, and it takes sound about 1ms to travel a foot!
Thank you! This one stuttered a bit, basically because (I think) I had the wrong thumbnail so a lot of people missed it. So thank you so much for watching!
Absolutely true. But, here’s the kicker, there was no midi. I tested all sorts of configurations, but the one that really confused me was when I just hit play on the drum machine and still got drift. Not only that, but some arrive early. It’s *definitely* drift at the drum machine/sequencer level.
@@DistortThePreamp I produced a lot of music in the late 80s, early 90s and used more equipment than I can remember. One of my drum machines was the R8 and I can confirm that it reacted differently when using its own sequencer compared to being triggered. Although both had issues. Everything has a delay/latency, keyboards have latency, sequencers etc. There’s even latency on different I/O of the same type. The period of time you are talking about “80s drums”, presumably the earlier half, would mostly have been CV/gate triggering for sequencing, which had its own latency issues, although, similar to MIDI in it being serial. Then you also have to consider the type of tape synch being used to synchronise CV/gate sequencers to tape, which was likely FSK, which would have had its own issues due to a relatively low resolution pulse. Which would have also suffered issues with its square (pulse) wave being rounded due to the many issues with tape. Even once we started using SMPTE to MTC, the issues of tape synch were still part of the overall latency equation. In short, the issue you are talking about is not unique to any time period of music creation with sequencers. These offsets you talk about where also part of the early 90s dance sound. Most of us doing it seriously would have been using something like a Midex on an ST to get multiple parallel outputs, which was only used to mitigate the timing issues as much as possible but there were still there. Then compound that with SysEX and CC data and the issues are still very much relevant today. Like I said, the timing errors you are talking about are not unique to the period of music you are talking about. Any “sound” differences from one decade to another are going to be primarily due to the recording and mixing technology and less about the actual sound making machines and their inherent issues. Or put it another way. If The Human League had have recorded and mixed Being Boiled or Empire State Human using the exact same synths but recorded and mixed them 10 years later, they would have sounded very different. Cheers, Dave.
Great video! Can I ask a question please? I have some synths and drum machines from the 80's. I'm a huge RX11 fan! I like to use either Cubase or Cubasis on my iPad to sequence them because I have noticed slight timing discrepancies that I like very much. You think this would be exaggerated by using the drum machines sequencer?
@@JamieMallender Interesting question. In my experience it’s different but not necessarily exaggerated. And to be honest all you really need is enough of a timing discrepancy that every combination of hits sounds ever so slightly different to the brain which, again in my experience, seems to happen before you actually notice definite timing errors. I suppose that’s a long way of saying I don’t think it would be exaggerated. But, even if it were exaggerated I don’t think it would be better. One thing you should absolutely experiment with though is tracking in the audio from the drum machine and then fitting the grid to whatever audio was captured (usually the kick drum) so then you have a daw grid but it’s following the crazy timing. I do this by creating a tempo map. This makes for an extremely flexible workflow and basically means that you should even be able to overdub drum patterns having ‘taken’ the drums. What drum machines and synths do you have btw?
@@DistortThePreamp That's very interesting. I shall definitely experiment with that, thank you. I have a TX7, Poly 800, FB-01, RX17, RX11, DR-550, a soundmaster SR88 and from the early 90's a DR-5 and a QY22. I've used some of these things in my videos. I like composing with a combination of hardware and software.
That’s a very interesting gear list. I wonder if you might like a rompler like the Proteus 1 to go with your FM and analogs? I don’t think romplers sound good enough to build a whole song around, but they can be fantastic for layers or supporting sounds. You basically use presets (with a bit of tweaking) and the obvious ones are the Emu Proteus, the Korg M1, or the Roland JV 1080 (or the 2080). It sort of depends on what kind of music you make, but I’ve got the first two and really wish I had a JV. As I said, I would never rely on them for a whole track, but if I want to reach for strings, brass, vibes, etc they are a great thing to try. Just a thought…
@@DistortThePreamp am I right in saying the Proteus is a rack mount? That could work for me. I don’t have room for any full size synths. My studio is full to bursting mostly with guitars and basses. But yeah, that sounds like it might be a solid purchase, thank you. ✌️✌️
9:33 Actually, that's true, but there is also a simple solution for the computer to bring more life into the tracks (especially to simulate this behavior). In Ableton, this is achieved by combining track delay and the echo effect with an LFO tool. Since the track delay cannot be automated, we use the delay instead. You insert a simple delay effect and switch it from beats to time. Then, set the effect to 100% wet and drag the time parameter all the way down to 1ms. Switch the mode to fade and Feedback to 0. Now, open an LFO and route it to the timing parameter, adjusting it to, for example, 8% and 10%. This will gently modulate the timing parameter, allowing us to mimic subtle timing inaccuracies. To make the modulation more random, set the waveform to Sample and Hold in the LFO tool (Random)-and voilà! It shift´s the track very sattle within 1ms etc. You can decide and play around with the LFOs speed settings.... This can also work in the negative direction. For this, drag the track delay parameter to for example -10ms and set the delay to 10ms. At this the tracks will be synced and play around with the LFO there :) Thank you for your video and if you have questions about it, you can simply ask. Best regards.
Thanks for this. Yea you’re absolutely right. The method I thought for Ableton is to use a neutral groove with a timing randomisation on the 16tg notes. But actually as a few people have noted the M4L humanise device might be even better. I’m not sure about an LFO because it would cycle (although in context that might be fine) although, as I’m writing this, doesn’t the Ableton LFO tool have some kind of S&H random setting? All good. And it goes without saying I was being a big flippant when I said computers were dead. What I really meant was that if all the transients are on the grid then that’s “dead”. But I know that even that depends on genre - deadmau5 is pretty grid based (although he advances his claps and snares) and his music doesn’t sound lifeless at all! Really appreciate you watching and taking the time to write this comment :)
@@DistortThePreamp Thanks for your work! Ableton's LFO has a random function, which can be found above the rate control where it normally says 'Sine.' Click on that and set it to 'Random,' which simulates Sample & Hold. I watched the video and thougt, this could be a possible solution and build it - it worked.
@@DistortThePreamp No, not at all. You have to switch it to time and turn the dial to full left, it will be at 1ms and from there you can adjust it to the right. If you need, you can hold shift and make very sattle changes with the mouse.
@@vladv5126 well that's exactly what this video describes though, random offsets from the grid. You just have to set it so that the offsets are no more than a couple of milliseconds.
Probably better to record the audio of the "tight" drummachine and then use effects to shift the microtiming, then to do it in MIDI, since MIDI notes can only be moved in form of ticks which isn't as fine as the audio grid or better: delay times.
@@mikosoft it's not quite the same. For some reason there's a pattern to the randomness of these old machines because they're actually trying to be consistent. The humanize feature in daws does the randomization all too well. It's actually best to just play the parts by hand with a controller and not quantise.
I’ve been thinking about this a lot. Here’s what I would do: 1. Make sure your drums (and other midi instruemnts) are all on separate tracks. 2. Load up a neutral 16th note groove from the groove pool that basically does nothing 3. Add some randomisation to the timing. You might have to experiment to make sure that the randomisation sticks to within about 6ms either way. 4. Apply this to each of the drums and instrument tracks. I haven’t tried this, but it should work. The important thing is that the instruments and drums drift AGAINST each other
@@DistortThePreamp thank you I will try this! A producer named Noisia who produces some of the best sounding drum and bass tracks stresses his preference for making things sound organic and in motion so I am excited to implement your info as I struggle making my sound feel alive (let alone sound good!). I love the 80’s sound, my doodles always gravitate towards that style, thank you for all the videos! Your Duran Duran arpeggio video was amazing.
Oh thank you! Yes, I think people focus on timing to make things sound organic, but overlook the microscopic sound changes (that are sometimes a byproduct also of timing). I’ve had so many people asking me the “how do I do this in Ableton?” question that I might take a deep breath and figure it out. I think it will work, but if it doesn’t then another possibility (maybe a last resort) would be for me to create and give away a whole load of 16th note grooves (taken from my actual setup) for people to use, but with the idea that no two instruments use the same groove. Hopefully it won’t come to that because that sounds like a rather over-elaborate solution. Do let me know how you get on though.
Nothing wrong with using a DAW, but instead of quantizing the recording you'd have to humanize it instead. And even if you'd quantize the MIDI-track for drums, the moment when you send that MIDI-signal to your Behringer RD-8 (TD-808-clone) and record the audio output, you will have these slight timing imperfections. Oh, and also think of the slight speed variation you'll have when you record with an analog multitrack tape recorder.
You’re of course 100% right about the DAW - all possible using humanise. But in fact there was no midi with the drum machine - it was just running free!
@@DistortThePreamp yeah, it just used an internal clock running at x bpm. And the rest (bass, guitar, keys) had to follow that pattern by ear, leading to more slight timing imperfections. This causes a natural groove and makes music come alive.
Also, weren't they all synched to an SMPTE track on the tape machine? Wouldn't that do more to account for drift than anything to do with the clock in the drum machine?
Whats the nature of the drift in the music you analyzed? Is it that each instrument is offset by a particular amount, or are they jittering so at any beat different instruments are triggering at randomly different times, or do rhe different instruments drift in and out of sync over the song like there's a slow LFO? I want to know more of your science!
Great question. They’re jittering so that the offset varies each time, including the direction. Rarely more than 7ms in either direction, but also never ever aligned. The fact that they’re never aligned leads me to believe that drum machines have some sort of serial architecture and there’s a minimum spacing of 1ms (similar to midi). It’s also random rather than an oscillator. For the experiments I ran the drum machine(s) on their own, and also as sound sources sequenced by external sequencers. I didn’t go into the detail in video, but tye timing got worse when external sequencers were throwing more notes at the machine. That seemed to matter less when the machine was internally programmed. All fascinating. I freaked out when I discovered all of this, but now I’ve decided that it’s part of my production workflow.
Interesting that external sequencing causes bigger jitter, and it's (apparently) proportional to the amount of notes. With some back of the envelope maths, MIDI is 32.1kbps, it's 10-bits per byte (with start & stop bits), and note on & note off are each 3 bytes, so sending or receiving a note or off will take almost exactly 1ms. Assuming everything is perfect and instant (when is it ever?), if your sequencer is sending a note-on message on three different channels on the same bus, the first and third note-on messages will be received 2ms apart. If in your sequence there are preceding notes ending right as the new ones start then for each note-on the sequence will first send a note-off and that 2ms doubles to 4ms. 4ms is equivalent to 250hz, so you're definitely in the realm of phasing. Very cool!
Yeah, this is exactly right. My suspicion is that internally the drum machines’ own sequences are implemented in a way that is either straightforward midi or similar to midi. Even the modern Behringer RD9 is incapable of spitting out hits that land together, and that’s when you press play on an internal sequence with no midi. Everything is always at least 1ms apart which I think is telling. And when I’ve got a huge sequence running for a song I might easily have 10-20 things happening on the same beat, albeit on different midi channels. I like the timing drift but that can get a bit too sloppy, so I often turn off the extraneous stuff when it comes time to track everything in, just to keep it a little tighter.
@@DistortThePreamp Wait, you're getting jitter? How are you measuring? I recorded my RX5 and then got reaper to slice the transients at a certain threshold and then I measured each slice. This method had the RX5 measuring accurate to with 50 samples at 44.1khz. Have you read the Roger Linn interview?
@@replaceablehead Yes I have read it. These were measured in Pro Tools, at 48kHz. Not only were there timing shifts but, much more importantly, the alignment between drum instruments changes. I’m not discounting Roger as a technician, but it’s worth point out that he has never acknowledged the effect that the inaccurate sampling in the Linns had on the effective timing. Because the samples weren’t sliced at the transients, the gap at the front varied as you detuned each drum. So literally every combination of drum tunings would have a slightly different feel. This was very pronounced indeed on the rimshot when it was severely detuned by, say, Prince.
Are hardware sequencers, like the Yamaha Motif xs also capable of creating that unique feel simply because they’re external from the computer? I often find that my inspiration and results are entirely different when I record from a hardware sequencer into my computer. Something about it not aligning perfectly with the grid brings a kind of magic I really enjoy. Is that the intention behind using hardware?
Interesting question. I haven’t tried the Motif but it very likely does the same thing. I suppose if it’s only sequencing internal sounds then it might not, but certainly if it’s sequencing external sounds it’s very likely to have the same “issue”. In terms of your question “Is that the intention…” I think that for me it’s one of the reasons I like using hardware so much. My original reason was all about the workflow - the physically of pressing buttons and turning knobs is very different to sitting in front of a computer monitor, and in my experience producers much better results, much faster. But the timing revelation opened up a whole new set of reasons to prefer the workflow. I can’t say it’s definitively better as there’s lots of ‘grid music’ that I love, but I think it’s definitely ‘better for me’, if that makes sense ;)
My best tracks are made by a mix of machine sequencing, hand playing and some sound processing synced up to one of the machines (that I somehow have to hand start). .
Just found this. Love it. You have a new subscriber. Now, gotto buy a stetoscope. Have a nice weekend! Best , old-school electronic musician ai accident
Oh fantastic! Make sure you get a vintage stethoscope - the modern ones don’t work ;) I’ve been literally editing all day (12 hours non stop) so this is a nice message to come out of the edit to 🙏
Thank's for that Video. You bring it on the Point. In my first experimental Recordings from 30 years ago I useed drum machines and sequenzed Parts triggered by the machines and it sound always intersting and human cause there is no masterclock by a computer. Today my Korg DDD 1 is the best tool to programming beats cause it has it's own heart and the sounds are as they are. 80ssssss. Love, peace and harmony Bernward
Old drum box had much more jittering like Roland TR 808 / 909; Linn LM1 / Linndrum. Modern clones completly lost this as they are clocked with modern processors, LM1 sequencer was driven by a VCO .
So this is the funny thing - I haven’t tested the modern Roland drum machine but my Behringer RD-9 absolutely has jitter! And from what I can tell so does my Squarp Pyramid. The latter is extremely hard to prove because there are so many other factors but I did some midi-free tests using the CV/gate and there was still drift. It’s all really astonishing. My best guess is that although they all have quartz clocks which in theory are extremely accurate, there’s some cheap chip being used to handle the timing. In truth quartz clocks were also used in the old units back in the 80s - the tech has always been there, it’s just that it was compromised by design choices (is my educated guess). For example, the Synclavier was only ever accurate to 10ms which is quite shocking since they marketed it as totally accurate. Etc etc etc. And of course the final irony is that in most cases the drift sounds good and makes the sound combinations vary.
Great info! I did a Podcast Ep a few years ago about playing on the grid. "Groove" can be a complex topic. I'll add this: as I Zoom in on the GB/Logic Drummer, the hits have "temporal phase drift". The Humanize dial in this case is about 5%. With it off, the beats are lined up with each other but slightly ahead of the Grid. The dial for me is a bit unpredictable but it's something to check out. Enjoying your channel.👊🥂
Oh that’s extremely interesting! And they’re not aligned, right? So the drift is on a per instrument basis? I think 5% sounds like a lot. I don’t know how the percentage is being calculated but you really only need a few ms of drift to get the effect. So glad you like the channel. Really appreciate the comment :)
@@DistortThePreamp That's right- Not Aligned. The Humanize dial does not give you a number (I wish it did but I'm guesstimating, based on a full turn) so I sometimes zoom to a fixed amount and eyeball the instrument displacement.
That’s a totally solid technique. There’s definitely a point when the brain can’t quite hear the timing differences but it can sense that the sounds are changing and the music comes ‘alive’. Very important IMO.
Small counter argument: sequencing hardware _from_ a computer will incur MIDI jitter. This might be a problem if you want really tight hi-hats but otherwise I say embrace it.
I remember reading an interview with BT (the dance music producer) maybe 15 years ago where he pointed out the same thing, but the opposite way round. He was a bit obsessive about making stuff perfectly sample accurate, laying it out as audio on the Logic timeline, because he'd noticed MIDI timing was all over the place and sounded dated.
Yes, this is what I used to do. We all used to talk about midi being ‘sloppy’ and we’re obsessed with being ‘sample accurate’. I’ve now gone the other way 😂
What is interesting about this is that he is right have you ever heard a drum machine miss a entire note and leave you with a very small piece of static fuzz that is barely perceptible. It won’t happen very often but when it does it will shock you… maybe
Yes! My R8 gets confused sometimes and drops snares. It would be hilarious apart from when I’m tracking and don’t notice. Then I have to give it a strong telling off.
I noticed timing variations in the 90s with MIDI. The track arrangement in the sequencer was such that first drums, then bass, then solo, then pads (these were delayed again) came one after the other, to get the groove somewhat tight. This led to a certain 90s dance vibe. I don't know the situation in the 80s. But I guess that in addition to MIDI, there were also sync fluctuations and fluctuations in the devices, which explains the pre- and post-delay.
That’s what I would have thought, but actually it’s baked into the drum machines. Those measurements were just from me programming a pattern and hitting play. No midi involved. Totally erratic.
@@DistortThePreamp Makes sense. Something that wouldn't have been on my radar at first. At second glance, of course, the processing speed of the commonly used 8-bit CPUs was also limited in order to fire samples perfectly on a grid.
Yes. It's particularly ironic with the Roland R8 drum machine -- that was promoted as having extremely high resolution. Instead of 32 notes per bar you could do 960 so you could, effectively, programme actual triplets and very detailed rhythms. Who knew that the actual timing error margin was way over this resolution. Not that it's not legit -- it's all legit. But, honestly, nobody knew, because nobody checked.
@@DistortThePreamp it would be really interesting if Roland knew that their older machines had timing inconsistencies, found the R8 to be too precise and re-introduced the inconsistencies so as not to lose the magic.
So what about the Fairlight CMI? It was basically a computer and was used quite a lot for making beats in the 80's (e.g. Opportunities by the Pet Shop Boys). It doesn't get more eighties than that...
Yes. Although there was still a surprising amount of analog midi gear involved in those records. Same with the Synclavier which had its own timing problems as well. It was marked as being entirely accurate, and it turned out what they meant was accurate to within 10ms.
Surely those timing errors are trivial to program into a modern DAW to recreate that 80s sound? Also, some audio examples to listen to would have been nice. I've seen your reply to someone else saying you're thinking of remaking this video with some samples, so looking forward to that.
I have a roland sh-01a (the modern recreation) which I send midi to from my DAW and then record the audio. I've noticed that the timings of percussive sounds vary quite considreably, by several milliseconds, and perhaps up to about 10. It's possible to see this variation in timing against the grid even when you have say 4 bars of audio across the screen. At first I was a bit perplexed by this and thought it must be coming from the midi as the synth is basically a VST in a box, or so I thought, but maybe this is Roland's way of making it sound a bit more analogue. Sometimes it does seem to add a certain something, but on other occasions it's downright annoying and I have to go in and manually adjust the timing. Another great vid, keep up the good work!
Well ai got it right. I was overlaying drum loops I generated with udio ai in the daw (mostly 70's and 80's dicso generated stuff) and noticed that my kicks with the exact bpm as the udio tracks don't align properly with the udio stems. Hense if your cpu and daw is crap you could also put vst effects that doesn't do anything on separate drum samples mixer tracks and they will introduce latency and will cause phase issues too.
It's funny how some people obsess these days about timing just like pitch perfection. I never thought there was anything wrong when I was slaving my Atari to my Tascam 244. I learnt back then to avoid quantization. However, the problem I had was eagerness to fix flubs in Cubase instead of practicing till I got things right - a very bad habit that I carried forward to digital audio multitracking that not only stunted my growth as a musician but resulted in a vicious cycle of increasingly lacklustre performances.
Oh wow! That’s the sort of sequencing cousin of my Roland R8 that was released around the same time. I don’t know it very well - I’m familiar with sequencing drums on the R8 which is a dark art, and I have a Roland MSQ-700 sequencer which is another dark art, but I bet it’s really good once you get to grips with it. Is that what you’ve found? You still use it!
Truth…? Let’s go anecdotal: Gerhards Lengeling’s precursors to Logic (e.g. Creator/Notator) ran on C-64 and Atari ST. Sync to tape (the only option, not phase perfect) was bound to recorded linear time code (e.g. SBX-80). A Linn 9000 synced to tape from a recorded square wave, counting pulses to BPM, maybe at 96 pulses per division. The Synclavier would use an even more expensive sync-unit with quartz PLL - but still the same tech as the above. All of this happened on complex “CPUs” less powerful than 4$ microcontrollers - Then, an 8-bit serial protocol (MIDI) is sequenced onto this heap of uncertainties…? That’s how we did this, at Polar in Stockholm or at Advision in London. Some lucky few, in their own studios. There’s no magic, just our silly grinding to get around the limitations of the best tools available for labels’ or publishers’ cash. As fun and frustrating as recording music has ever been. Learn your tools (that’s what they are). The magic is within yourselves, as musicians interacting - not inside retro boxes of unobtainium, barely able to crunch serial data… 😅👍
i like to compose midi on the computer sequencer then record the midi to the hardware sequencer for recording or live performance, the clock on these machines have a vibe as you say, i tried working in the box for years i could never get that timing/feel, every hardware sequencer has it's own feel going on.
@@DistortThePreamp sounds tedious but i found it the best workflow to get the results i want, none of my harware sequencers have screens so edits at the note level can be a pain, although i do like using my step sequencer to jot out ideas, ti like the computer sequencer for are editing (prog changes, cc data, sysex) and saving work, then i record that to the machines, works a treat. : )
I've messed with using random midi effects in Cubase to do this. Bit of timing and velocity variance on a sequenced pattern goes a long way. I mostly use it for arpeggiated piano though, like others have said it would never occur to me to do it for an 80's song.
Great video love the detailed explanation and the mysterious questions haha. I use Fl studio and I know on each drum sample has an option to let you offset the sample. Would that be a good way to emulate the slight time differences you spoke on? I
Many don't realize that there's an inherent delay in MIDI 1.0 to hardware. A single note-on messages takes apx 1ms to send over the wire, assuming it's the only thing being transmitted. If you use MIDI like we did in the 80s, using MIDI Thru on devices, you can get even slightly more variation. But more importantly, when you have multiple channels of MIDI data (multiple synths) then you end up with even more potential delay between note messages (up to 15ms more delay for a note on if you manage to max everything out).
So even without the delays introduced by underpowered CPUs of the day, with MIDI, you already have some variable trigger delay in there. Of course, this only applies to MIDI, so post-1983 stuff.
Yes absolutely. What was especially strange about these tests though was that the random drift continued even when I took midi out of the equation and just pressed play on the drum machine. Also, I started to guess that the way the drum machines were implemented internally was actually serial (like midi) because nothing *ever* lined up and was always at least 1ms out.
I was wondering what caused the timing issue, I was thinking along the lines of US instruments being converted for UK 50Hz electricity.
My set up must have some serious timing discrepancies as my MIDI set ups are a rats nest of cables.
I wonder why MIDI never evolved into an optical system? probably cost!
@@AndrewAHayes Yeah. The original MIDI standard was about using inexpensive components that were super easy to integrate in 1983 because the CPUs all had serial integrated. Optical is more expensive and requires many more components. MIDI 2.0 is faster, and with the microcontrollers available today and used in synths, network MIDI 2.0 (spec now complete and being voted on) has a ton of promise for being the inexpensive super fast solution (also without the distance limits of USB)
@@Psychlist1972 Still no chords though, right? And you still get ‘gapping’ with the cc data because it’s relegated below the notes. People always say that they didn’t expect to be listening to suchandsuch a song forty years later. Well I certainly didn’t expect to be peering over tiny midi monitoring screens forty years later, trying to figure out if a channel is properly taking priority 😂😂😂
@@DistortThePreamp the MIDI protocol itself doesn't have any prioritization other than system real-time messages like clock. Everything else transmits as the sender wishes it to.
Chords in MIDI 1.0 are serial, so if you somehow managed to hit 5 notes at the exact same time, there'd still be apx 5ms between the first and last, assuming no other traffic.
In MIDI 2.0, they are still serial messages, but the timing is in the microseconds, not milliseconds so it becomes far less of a consideration.
An example would have been nice. You stated you tried it by recording it "normally," and then quantized(?) it, and there was a noticeable difference. Well, why didn't you share that with us?
Yes, without A/B testing then it's just a hypothesis. It's like "trust me, this is why". All music up until the 2000s was using crappy MIDI drifting timing, so I don't buy that is what defined the 80s sound, at all.
Yes. I’m really sorry about that. I made a previous version of the video which contained the full experiment but I couldn’t get it below 35 minutes and so, perhaps wrongly, I decided to remake the whole thing in a simpler way.
The original experiments were done a couple of years ago, but the new experiments I did and filmed for the video worked as follows:
- I sampled some of my drum machine sounds into the DAW
- I programmed an identical pattern in (a) a midi clip in the DAW, and (b) my sequencer. (Note: I should also have done it in the drum machine which I have previously tested).
- I recorded both.
- Switching between the two records they sound the same, but they’re not the same as I then prove:
- I take the kick-snare audio sum from both the midi audio and sequencer audio
- For each of those I duplicate, layer with an offset of two beats so every kick-snare is under the previous kick-snare etc
- I then flip the phase of the duplicate and show how with the midi audio it produces silence (a successful null test) and with the sequencer audio you get a phasey hollow sound that is noticeably different with every hit.
- This proves that they’re different, and also different every time. And is also a perfect explanation for why DAW beats sound different to beats that were made in an old fashioned way using drum machines or sequencers.
- I also reference some science which explained how, although humans typically perceive sounds that occur within 5ms of each other as occurring at the same time, our brains are highly attuned, albeit subconsciously, to the sound of phase shift because it’s how we perceive front to back, and from an evolutionary perspective it was an important part of staying alive when there were predators behind you.
- Thus, in a slightly humorous way, one might say we have evolved to be able to distinguish between DAW beats and drum machine/sequenced beats, even if our modern brains can’t explain what exactly we’re hearing that’s different.
- Finally I note that for people with proper training, they can pick out the differences and demonstrate with a blind A/B test. This last sequence was planned and demoed but not filmed.
I’ve done a lot of thinking about this, and I am coming to the conclusion that I may entirely re-make this video in a few months with more of the science and demonstration.
In the meantime, sorry for the hand waving.
This was a suggested video for me, and I’m quite interested. I appreciate the thoughtful response, so I’ve gone ahead and subscribed so I can see the follow-up. Greatly anticipated.
@@DistortThePreampyeah, you definitely need to provide examples of your findings here!
because he's just repeating the same thing all the yootoob experts have been saying five billion times...if you showed and a b the video would lose its impact
This video surprised me, namely that apparently so many people are unaware of midi humanizers. The M4L humanizer I've been using for about 10 years which was based on statistical findings of real human drumming published in studies like "Synchronization in human musical rhythms and mutually interacting complex systems".
Well this is true. I alluded to humanisers, right? (Or did I cut that bit?) I can’t remember whether a M4L humaniser can advance as well as delay midi, but if it can’t than the obvious workaround is to use track delay to advance the track and then double the amount of humanisation. Although as I’m writing this, it probably doesn’t matter that much - the main point IMO is capturing not so much the timing drift (which is what humanisers were intended to do) but rather the way the phase relationship changes between two overlayed sounds as a result of a differing drift in thei alignment. Great point.
So the timing feeling humanly natural doesn't matter as much as each drum hit having some unique quality caused by phase "issues". This is, i imagine, why people still refer to this era of music as rigid.
EZDrummer has this built in too. the funny thing is they also use real drummers and people sometimes it humanize that and it stuffs the sound up as it kind of doubles the effect makes it sound unreal
Ive always imagined this was the case - I first noticed it with the Linn Drum where there is a noticeable lag if you put two sounds on the same subdivision - only one sounds in time, the other one comes in late. If you listen to those early Prince albums using the Linn Drum the Kick is almost always fractionally late on the one. He then took that on as a key part of his sound with his live bands always dragging the one - its part of what makes his feel so funky. When I first got into sequencing I used a Roland MC500 which seemed to have a great feel. When I switched over to using Vision on an early Mac laptop it still seemed to but when I switched to a modern DAW and had sounds actually IN the computer rather than on an external machine, I found that if I quantised it was too mechanical - I had to deliberately leave in timing errors, especially on drum programming, to get a natural feel. It's brilliant that you've taken the time to actually create a comparison and get the evidence! It would be cool to hear its
Exactly. There’s another reason that’s peculiar to the Linn which is this: all of the samples had ‘dead air’ in front of them. There are lots of theories as to why this would be the case, but my guess is that Roger Linn wanted to make sure that the VCA would have time to open. However, what’s interesting is that as you define the samples the dead air (of course) gets longer. This means that the groove slightly changes depending on the drum tuning. And Prince often used very detuned claps and rimshots. I am fascinated to see/hear if the Behringer clone replicates the dead air. If it doesn’t, it won’t capture the Linn groove.
@@DistortThePreamp I didn’t know about the dead air, but that makes a lot of sense as to why those detuned sounds feel so late! I wonder if he was de-tuning primarily to get the timing effect and accepting the pitch shift as the side effect!!
didn't prince play the linn parts himself, without sequencing?
@@CodyHazelleMusic same thing still happens. If you press two buttons exactly together, one will sound first!
The phase argument also makes me think about the oscillators in analog synths that just keep running, instead of restarting with every 'note on' like in many (but not all) digital synths.
That’s a very good point indeed!
your channel is such an absolute goldmine, thank you so much
I very much appreciate this! Thank you so much for the comment 🙏🙏🙏
A lot of the timing in discrepancies were also due to the fact that they recorded onto 2 inch tape
and a lot of cut and splice took place inherently changing timings of different tracks
Which, when it comes to tracks from prints, and those classic 80s anthems clearly added to the complexity of the finish product
But as a user of hardware, I can agree that when you look at it, after recording, you can clearly see it’s about the groove
I love your analysis ❤
Yeah, I was so surprised about this. I had assumed, like you're alluding to, that the timing discrepancies were to do with either tape stretching, or general clock drift. I was astonished to find that the drum instruments were drifting independently even when you just hit play, in other words with no midi involved at all. So when you add in SMPTE clock, etc etc etc, it all sounds very non-robotic.
Makes sense. The spec. for the MIDI 1.0 protocol guarantees a maximum latency - “response time” - of 8.3ms (at 30 frames per second). That's well within what would be comfortable for live performance, but it's not nothing. And it fits rather neatly with your observations of around 7ms of drift.
It never occurred to me how it would affect the sound. But of course it will.
Oh I didn't know that! And of course 8.3ms = 16.6 maximum delta. I was aware that every midi note had to be at least 1ms apart, and there is no such thing as a midi chord (because it's a serial protocol), but I should obviously have read the spec. Duh.
Interestingly the Juno 106's voice CPU has a "cycle time" in which it updates all six voices and the "global" parameters (noise, master VCA, PWM, resonance, and the various switches) is 4.3ms, but by the time the assigner CPU has done its work and it's told the voice board over its MIDI-like internal link, it averages out at about 6ms latency between "Note On" message and sound coming out.
I'm still curious about recommendations for a modern hardware sequencer, however... this morning I wrote a script for Reaper that will randomize the placement of all selected items by up to a certain number of milliseconds, plus or minus. I use Reaper for my work so it's the DAW I generally noodle around in with music.
I created a drum and kick track using an Oberheim DMX plugin, printed them to audio, which resulted in them all being perfectly on the grid. I then used strip silence to isolate all of the hits, then ran my script and entered 7ms as the value. It randomly pushed and pulled all the individual drum hits by up to 7 ms forward or backward. It definitely changed the feel of things.
I think purely random may be the wrong approach, as it sounded a little chaotic. Maybe the better way is to, for example, pull all the kicks forward by 3ms, then have it push them forward or backward by up to 1ms more after that. To make it a little less random. I'm definitely going to play with it though.
So you sound like you’re properly mathsy, and therefore I don’t think it would be a random linear deviation around the mean. You might want to try a power distribution with an upper bound, but there’s another thing which is that none of the hits ever occur together and are always at least 1ms apart. That’s probably because the internal drum machine sequencer is modelled on midi and is serial. All lots of fun.
Regarding hardware sequencers, I absolutely LOVE my Squarp Pyramid! Unfortunately they’ve retired it and are trying to move to their next gen model which looks great but is a bit physically big for me. So my recommendation is a second hand Pyramid. At some stage I’ll make videos showing my workflow, but suffice to say it works excellently for electronic pop and rock.
Hit me up on my email in the description if you have any specific questions…
I’ve worked as a sound designer in the video game industry for almost 20 years. I take advantage of this phenomenon all the time when making transient sound effects. What I didn’t know was that this was happening all over 80s albums but it makes perfect sense.
Oh interesting! So you are taking identical samples and exploiting the fact that when they layer the sum sounds different because of phase differences? That's fascinating! And let me return the comment -- now I think about it, it makes perfect sense in your industry too...
@@DistortThePreampSometimes identical samples, sometimes similar samples. Sometimes I’ll cut off the first one once the second one kicks in, kind of like “choking” it in drum machine terms.
We do this for explosives, gunshots, or anything big and transient all the time. Theres a huge variety of sound you can achieve when playing with very short amounts of time offsets. I remember when I first learned this, a more experienced sound designer taught me and it absolutely blew my mind how much a few ms could change the timbre of the sound.
Oh that sound designer would have been talking about exactly the same thing I was. It’s extraordinary once you witness it with your own ears. I’ve never done any game sound design but it sounds fascinating. Your description of ‘choke groups’ is obviously very applicable to music…
Fun. I drag my tracks in the daw a few ms to the front or back and then repeat till i get what i like most. But i should repeat it more later in the track 😎
This is cool to read
Fantastic. Thank you. I was the Fairlight Series III CMI demonstrator at Peter Gabriel's 'Syco Systems'. The timing 'issues' took me to Australia and back, then we made different versions of the software for different clients, specifically due to the quirkiness of the Series IIx and Series III. The NED Synclavier was within about 1 ms, however many of my clients preferred the Fairlight 'timing', as Trevor mentions in his book. Here we are still working with the original Fairlights and I use them for specific jobs for making soundtracks / music productions - still! :-) Your 'Temporal Phase Shift' is great. Another important element is that the Fairlight had separate individual outputs 8 and 16. Therefore the calibration of each would be slightly different. So each individual note would be both temporarily 'phase' shifted, as well as the frequency / EQ curve slightly different on EVERY note, depending on how many output (or Nphony) was assigned to each multi-sample. If that makes sense?! In other words, over 50% of every hit record from that time, needed the Fairlight somewhere in the production to make it sound like the 80s. Emu and Linn also had their own quirks, as well as the Ensoniq Mirage. I hope this helps? Cheers! PS With the Fairlight 'Page R' or 'RS Page' timing priority was given to the top 'track' and became more 'wobbly' the further down the sequencer page. There are stories that I could tell..........
Roger - such an interesting comment!
I had no idea about the Page R priority! That’s hilarious! What was surprising to me was that I could hear the alignment shift in the sound combinations well before the timing errors were apparent. I was a bit coy in the video (although I dropped a very large hint) but that was an actual multitrack (that I shouldn’t have) that I was showing of a famous number one produced by Julian (whom I’m scheduled to interview in a few weeks btw). You can literally see the drift. Some of the snares are ahead, some behind, etc etc. I think everyone has historically focussed on timing discrepancies as a rhythm issue, but of course it’s just as profound if not more profound as an alignment issue between sounds that makes every combination sound different. I’ve just been doing some snare sound design where I’m layering two identical samples and delaying one of them. As you would expect a delay of only 1ms produces a very noticeable comb filter.
The Linn timing issues are fascinating. To me the biggest ‘secret’ about the Linn is that there was dead air at the beginning of the samples which of course got longer when the samples were detuned. So when Prince was detuning his rimshots he was actually changing the groove. And that’s before one analyses how accurate its internal sequencing clock was.
If you feel like it, get in touch over email (address in the description). I have too many questions for this short reply!
@@DistortThePreamp Great. I'm in touch still with most of my old clients as there's 3-part documentary in production at the moment. I'm happy to chat via email :-) I 'curate' a number of old machines...
I've been recording an 80's disco/goth fusion EP, and i made this discovery myself recently. I'm sequencing on a 606 as i love the workflow.
Recording a bar at a time, and after I'm done I'll adjust the daw's grid to match what I've recorded. The looseness of the timing is an interesting thing. Apparently the 606 actually has capacitors in circuit specifically for this, not randomness, but looseness.
This video really did put my mind at ease, i was working around all these inconsistencies, my ears were telling me it's fine but my eyes were not.
You know what they didn't have back in the 80s? Graphical waveform readouts.
Exactly - no graphical waveform readouts 😂 That’s why I wonder whether I’m one of the few people to ever have done these tests. But I’ve decided that I very much like the sound of the drum machines.
Pro tip: a red stethoscope is fine for 80s music but to properly calibrate a 70s vibe it needs to be green with yellow flowers. And carved from hazel.
That’s absolutely correct. I find the hazel responds better to the different frequency gamut.
Agree, I have two.
Reason studios has a drum module called “Umpf Retro Beats” that has vintage 80’s drums and the gated reverb. It also has a “regroove mixer” that you can put your drums on different grooves and you can “slide” the midi so you can get this drifting effect.
Congratulations for this video. It really is helpful and we all can see you have a beautiful passion when speaking about the 80s music and overall sound. Thanks.
Thank you so much! That’s very kind!
“The absence of limitations is the enemy of art”
Well put.
Totally agree. All the best electronic musical innovators produced their best stuff when they were forced to work within the limits of technology.
@@andygriffith5160 💯
Loved your obscure PSB's reference and for letting me know the song.
This and the NO code made me subribe immediately. As a producer, this stuff is as addictive as crack.
Huge compliment! Thank you!!!
I’ve been using an Akai MPC60 mk1 as my main midi sequencer for about 15 years now with a studio full of synths, samplers and midi kit. It’s tighter than any computer based midi sequencer. Even better than my old Atari St from 30 years ago. The Atari was tighter than any Mac or PC based sequencer but the MPC has the edge. You can offset the timing on the MPC too. It gives such an organic feel. I’ll then multitrack the audio into the computer. Slaving the MPC (can I say that word now?!) to the computer via midi clock.
I do a similar thing - I start by programming in my sequencers, then I track in and slave back to the sequencers using compensated beat clock. It’s all right enough. But nowhere near as tight as putting stuff on the grid inside a daw and hitting the render button. And this IMO is a good thing.
@@DistortThePreamp what does "track in and slave back to the sequencers using compensated beat clock" mean?
Ok. My mind is blown once again. I never thought they weren't perfectly aligned.
I was stunned when I discovered this. As I said in the video, I spent about a month trying to solve the ‘problem’ till I did, and then realised it wasn’t actually a problem. I honestly don’t think even the 80s producers know this. I was out with someone who used to work at Trevor Horn’s studio with all of the usual suspects, and it was certainly news to him.
That also blew my mind
I just noticed this yesterday myself when I set my DAW to 140 BPM, the drum machine synced up nicely, but sometimes it dropped to 139 BPM for a fraction of a second then jump back to 140. It happens relatively often.
Creating 80's megamixes though is a nightmare for that very same reason.. I have to warp the tracks in most cases every couple of bars and then offset afterwards once I have introduced newer drums to bring back that live feel. It's funny when you think about it. When there were imperfections in beats, everyone would clamber for perfection. Now we have it? Look what we want to go back to. (I say we, because I am guilty of it too)
I know, right! The reason I got Ableton 5 back in the day was solely to warp tracks and put them on the grid. And now here we are.
New fave channel!
My personal obsession - The wonderful processing of vocals for 50s/60s country music such as Marty Robbins
Oh thank you! Old fashioned vocal processing, particularly country, is something of a lost art. Technical limitations made it impossible to do the kind of processing and overdubs that became standard even by the 70s. So if you wanted to change the EQ, you’d better select a different mic. Similarly, rather than rely on compression, the recording engineer would learn the song and literally ride the vocal fader (that sets the recording level) during the singer’s performance. It was a completely different world.
Since 30 years ago not only we always loved the famous MPC feel, knowing it’s swing factors, and realizing the MIDI transmission imperfections like you very well mentioned, but I also noticed the various synths had different speeds to receive busy MIDI info and to generate the corresponding sounds. And it doesn’t stop there… the different sequencing softwares had different timing characteristics! I remember loving the feel of the Opcode’s Vision and I used it on many many records, until it was sold to Gibson who sadly killed it. I never really understood why they bought it! Anyways, after making records all my life, I rarely get impressed with what I see people saying, but you truly are a genius. Your videos are very unique, intelligent and inspiring!
Thank you! The MPC is really quite special sounding. Really great workflow too that doesn’t get in the way of creativity. I don’t know much about the new ones - it’s potentially a bit over-complicated - but there are still a lot of great records being made on the 2000XL.
I remember recording on my computer back in the day using an E-mu Emax II and an Alesis MMT-8, and I never quantized my music. When I recorded my first album in the 2000s, the trend was completely different-it was all about achieving a perfect sound. Projects like Thievery Corporation, Jazzanova, and those on the Naked Music label were focused on sounding flawless, so everything was quantized. I decided to mix both natural instruments and quantized elements to give the music a more 'alive' feel.
Have a look at the “Human” vst by Mixed in Key . I enjoy setting instrument part variations in my DAW (+ or -) 5-7ms per drum instrument part and then add Human to get the variation in timing. Saves a ton of headroom and adds that secret sauce to keep your ears interested.
DAMN your channel is badass! I saw the title and thought "I imagine it's gated reverb? I'll be back to watch this soon" but this is SO much more than I could have imagined 🔥🙌
Oh fantastic! It took me a while to get the title of this right. And you wouldn’t believe the pushback I’ve had - it’s easily my most co traversal video (and I really never try to be controversial as I think it’s bad entertainment). So I’m SO GLAD you liked it :)
Back in the early 90s when i was studying audio engineering one of my teachers told me: record your drums by hand and loop or copy paste each 4 measures, but do not quantize. Or you can quantize the kick but do not qualitize anything else... duplicate and loop but do not quantize, you need those harmonics around so it can sound organic.
@@DF-eg8vl This is exactly right. Very good approach!
What a great video. You have unique talent in explaining thngs. I always assumed that the majority of 80's drum machine parts were either programmed in the machine, or alternatively quantized after being played manually, but I later found out that artists like Michael Jackson's musicians had the Linn Drum parts played by hand and not quantized afterwards. Other bands played live drums on top of a programmed drum machine (Fade to Grey etc). I would have thought that drum machines have their own rigid clock like on a DAW sequencer, but not so! There is a temporal phase drift as you correctly mention!
Thank you! I was as surprised as you that the clocks weren’t and aren’t rigid. From what I can gather it’s simply a function of an underspec’d processor because the underlying quartz timing is very accurate. So the question becomes why on earth would they underspec the clock? And I’m guessing the answer is that they felt the timing was ‘good enough’ and didn’t want to spend the extra money. I’m not complaining, but it’s pretty hilarious. And it’s even funnier that my modern Squarp Pyramid sequencer keeps time like a drunk Uncle at a wedding.
Would have been so great to hear a comparison of the “hardware” sound vs computer sound so I could actually know what the special timing drift actually sounded like. I thought for sure that since you had the isolated tracks and could probably replicate the same beat on a computer, that at the end you would have played it back to back. Interesting video for sure!
Yeah, I’m really sorry about this. Loads of people have made this point. It was my mistake really - I had done a version of the video with exactly that, but it was over 30 minutes and I was worried that it was too boring. I will re-make this video in a few months in a different way and show the full experiment :)
@@DistortThePreamp that’s good to hear, looking forward to it!
Me too!
Beautifully put! Get out of the box for sure!
Absolutely! Get out of the box!
Dear Sir,
Please accept my apologies for my previous comments. I have watched your video again and realise my late night brain was misunderstanding the points you were making. You have explained and illustrated very eloquently and succinctly a huge amount of technical and creative information that helps us to understand what makes 80's music so great. As a content producer myself (not on this platform, and not music) I enjoy the indulgence of a little imperfection here and there.
I was a young teenager in the 80's, and often felt that some of the pop songs were a bit "thin", but when I listen to them now (in my 50's), I realise that they have a complexity and sophistication way beyond what I was able to appreciate at the time. I did enjoy much of the music that was generated in that era, including many of the bands that you reference in your videos.
I appreciate your wisdom, and have accordingly subscribed and enjoyed a number of your other films. please continue to share your insights with the world.
Kind regards.
PS, as a young teenager I was involved in a community "radio station", which actually involved cutting and splicing tape. later, in video editing, I came to understand that you could take advantage of the natural softening of a (electronic, not physical) "cut" on magnetic tape to make an audio cut more natural. I think that relates more to part 2 of your film.
Thanks again.
Dear Tony, thank you for these very kind words! Your experience of cutting and splicing tape is interesting. Were you doing something like deck-to-deck editing on U-matics, or Betacam, or something like that?
@@DistortThePreamp The cutting and splicing of tape refers to when I was involved with a school "radio" station when I was around 13 yr old. We did a 20-30 minute programme broadcast over the school tannoy system on a Friday lunchtime. Music requests, shouts, news etc. So it was 1/4 inch audio tape. we would use iron filings on the tape to get the exact cut point on the signal path. Razor blade and splice block for the (diagonal, for the cross fade) cut, sticky tape to join the bits back together. long time ago, so I don't really remember a lot of the details. We'd probably call it a podcast these days.
The school also had a Black and white 3 camera TV studio that I was involved in a couple of projects. Probably what sparked my interest in video, though it would be around 20 years before I got to that in a professional sense. It was all analogue at the time. I remember one of the techs that helped run the studio showed me that if you set the vision mixing desk up in a certain way, you could use the cross-fade control as a "threshold" for a cut, and if you got the sweet spot, it would split the screen vertically with a random zig-zag pattern. That blew my mind. And was very much "a la mode"
The video thing I mentioned was Betamax SP. Same tapes that run on Betamax domestic players, but at twice the speed for broadcast. not really possible to physically "cut" that in the same way as audio, as (I'm sure you know) the video track ran on a slightly curved diagonal across most of the track, while the audio tracks ran in parallel along the side.
So, to come to your main question, in my early days as an assistant editor (video), I was trying to do a really tight audio edit on a Betamax SP tape by setting up the audio mixer to do a one frame (25fps) cross fade, matching the vision cut, triggered by the edit controller. But the audio mixer couldn't do a cross fade in less than 3 frames. The supervising editor said "just make a cut ! ". with some disbelief, I tried it and it came out as a perfect soft mix. After some reflection (and I still marvel about this to this day) I realise that the malleability of the physical structure of the materials embedded in the tape would only partially move, creating the 1-frame dissolve at an atomic level. Crazy nuts stuff when I think back on it !
PS, I hope you don't mind, I have deleted my negative comment from a few days ago. Kind regards.
T
I can’t replicate the results. For my first experiment I used Reaper's “Dynamic Split Item” feature to split up the transients by threshold on a recording straight out of my RX5. It was accurate to within 50 samples. Not satisfied, I tried manually, with a similar result. Then I thought about how the 606 has a reputation for inaccuracy so I recorded some of my Behringer RD-6, amazingly it was even more precise.
Here are the results in milliseconds I got by manually slicing at the sample level.
For the RX5: 0:00.249 0:00.248 0:00.249 0:00.249 0:00.248
For the RD-6: 0:00.344 0:00.344 0:00.344 0:00.344 0:00.344
That’s right, the inaccuracies can’t even be measured in milliseconds with me doing manual slicing. Am I misreading the Reaper figures, are they not milliseconds? I’m so confused.
world class video on internal clock drift, thanks for making
Haha - thank you! I was so staggered that it even existed. I mean they’ve all got quartz clocks. But turns out the clocks for each instrument drift independently anyway. And the irony is, it sounds fantastic.
@@DistortThePreamp delighted with your reply thanks peter
Am about 75% through, you've still not revealed your insight (predict the internal clock of the sequences to blame), and you've presented a solution to an issue I was having to remove the machine like exactness of the modern computer. Slower clock speeds. And here was me trying to programmatically add a random drift of 1% to my tempo.
Thank you 😊 🎉
Oh fantastic! I still don’t really know why all the internal clocks are so bad - not that I mind, because I far prefer the sound. But my quest continues…
Great video, and I think you're right!. I'm curious about your recording workflow... assuming you record into a DAW, do you sequence everything using your external Pyramid sequencer and then record it all in one go? Or do you sync the sequencer to the DAW which would then allow you to do as many takes as you want? I have a Roland MSQ 700 which I can sync to my DAW for start/stop. This would allow to me program a sequence in the MSQ 700, sync the MSQ 700 to the DAW, record the sequence, program a new sequence, then record it, and have it all line up in the DAW. Is that what you recommend or would syncing to the DAW "fix" the "problem" and ruin the effect?
That’s pretty much it. I’m going to make a full video at some stage, but basically I start by tracking in the programmed parts from my pyramid into Pro Tools, then I do a tempo map in Pro Tools based on the kick drum, then I switch the master clock to Pro Tools so now the DAW is driving the clock. This means that if I need to do a sequenced overdub I can and everything lines up. I’m very familiar with this workflow in Pro Tools but I never really got it to work properly in Ableton. I am guessing that it would be possible in Logic and the other main DAWs…
The physical distance that MIDI data AND audio signal travels over the various wires between all hardware devices will also create its own natural lag, however minuscule, providing a nice sense of "timing" fluctuation as well. All clocks are a little different. I used to perform my live sets (DAWless) back around 1997-2003, like a DJ would spin records. No midi connecting any devices, only two Roland Synths (XP-80 and MC-505), and my natural sense of timing for triggering the starts of my next songs. It had the benefit of sounding amazing, or going horribly wrong, but it was human!!! It makes all the difference. 🙂
Yes. Audio cable lengths are not all that significant becasuse they’re traveling at the speed of light, but audio through the air is obviously about 1ms per foot, and midi messages are spaced at least 1ms apart and, given that midi can’t accommodate chords) this means that a four note chord contains a spread of at least 3ms. As you say, this can add up to quite significant alignment issues that simply don’t exist if you programme inside a DAW. It just sounds different.
That was extraordinary! Well done.
Thank you so much!
I’ve noticed this on extended 12 inch mixes where there’s lots of muted tracks and solos, you can hear subtle “mistakes” that make it feel more human even though it’s electronic, I used to put it down to processing power being a bit slow with tiny lags
You’re absolutely right about 12” mixes from the 80s. Not to descend into a history lesson, but the commercial reason they were made is that the chart rules recognised multiple sales (up to a limit) as separate units sold. So if you could persuade a punter to buy the 7” and 12” then you would notch up two sales. Therefore there was huge pressure for 12” versions to be produced of everything and very little budget to do them properly. I know several producers who used to make them in an afternoon just by basic razor blade splicing. They would mute the vocal and let the mix play, then add a middle bit, then repeat from the beginning with the vocal added. Or something easy like that. I know Tape Ops who were told to “do the edits” because the producer couldn’t be bothered, and sometimes the edits weren’t fantastic. “Shake The Disease” (Depeche Mode) has a very bad edit in the middle of the 12” version that I always found annoying.
Heart rate is subject to frequency variability (at rest) without our feeling or knowing it, this is an awesome convergence of living systems.
I think this is exactly right. Very interesting point.
Trivial to add a bit of randomness in DAW drum tracks - generally called 'Humanise' or something similar. And the delay between voices in a hardware synth is not random - any delay between the first audio 'attack' of each different sample relative to a different one is constant for every beat. They do not drift around.
In Logic some native plugins like the Drummer, have a humanizer function. That helps and there is a free Plugin from Baby Audio, named pitch drift, that helps to get rid from the steril sound of samples.
Yes pitch drift is great
Recently I have been listening to early Soft-Cell, Ministry and Depeche Mode, and was thinking how organic and lively most of that music sounds, and even thought it’s mostly drum machines, synths and sequencers, you can tell that it’s not hard quantised at all, to an extent that it almost feels like those instruments were played in real time by humans because some of those sounds really have a groove. Now I know why!
Some stuff might / will have been overdubbed, played by hands, onto the sequenced basics. Example: Basic Drum Pattern sequenced. -> Cymbals and tomtoms "overdubbed", by pushing the buttons by hand.
Most DAW's have a randomise timing option. I use Mulab and that has the option. My Roland W30 sequencer had it too. It's a time old trick.
Great video, really interesting! Just to play devils advocate, the latest version of ableton live allows you to 'loosen' the quantized notes by a random amount within a specified range. Would that not have the same effect if you programmed your drums in a piano roll? In Cubase there is an option to similarly randomise velocity with a range which i use on things like pianos just to give them a bit more of an organic feel as i can't play well at all so usually rely on hard quantizing, then using this to add some humanity back in :)
Yes, it basically would! I was overstating for comic effect when I said the computers were dead - you can absolutely do what you’re suggesting, and in any case ‘grid music’ can sound fantastic! Having said that, I truly believe that this sound (ie the way the sound combinations change because of micro-alignment issues) is fundamental to why 80s music sounds the way it does. By the time the 90s came along it had been ‘fixed’. And the issue didn’t exist in the 70s.
Great vid. Always interesting stuff going on in your vids.
Thank you! I am getting sooooo much grief, and rightly so, for not including examples. Ironically I filmed them but ditched them because I thought they were too boring. That’ll teach me.
I tried measuring my Yamaha RX5 in reaper, it's accurate to within 50 samples at 44.1 kHz. I also saw an interview with Roger Linn where he said the groove in his drum machines is a myth and they are dead accurate, although the MPC does have swing. From what I'm reading only 606, and 808 have any drift and obviously the older Roland boxes from the 70s. What I want to know is when people were using things like the BBC/UMI sequencer what was setting the clock? And further, how often did midi jitter come into play?
Great video. I found this out early on in my production. I use a sample delay to slightly offset my drum hits.
Fantastic! Great technique :)
@@DistortThePreamp thanks for sharing
One would think that digital audio workstations would have an option/plug-in to introduce random tiny timing imperfections to the individual rhythm tracks in order to give it that feel.
For sure, especially with the Synthpop in the mid-80's, Freestyle and very early House and Hiphop, there was this sort of "drifting" that's very "80's". I didn't know it was early midi, I thought it was just tape or something to do with tracking but okay, that totally makes sense
Thanks for the insight, it's definitely heard but awesome you brought it forward from the subconscious, thank you so much for the talk
You’re very welcome! Thanks for watching and for the comment :)
very interesting video! i love vintage gear, but most DAW’s have a “humanize” feature that will anti-quantize midi notes by introducing a little random variance. no reason to reach for outboard sequencers unless that work flow is genuinely preferred or some other juice is being sought after
That's because it was all still recorded on tape. Tape varies in speed as it spools forward > hence the out of time phasing. :)
Ah, that’s a good theory, but actually that’s not it. That could in theory produce a drift, but not an alignment problem. In other words the instruments wouldn’t move *against* each other. Also tape running at 30ips (or even 15ips) doesn’t drift that much. Meanwhile, one of the old Fairlight team has actually confirmed that there were indeed timing errors! Really appreciate you watching though, and taking the time to comment :)
@@DistortThePreamp Thank you. I was speaking from experience. I tried to remaster something off 4 track cassette and recorded each track separately into my DAW and they don't line up. Also, sequencers back then did drift as well when using CV instead of Midi which was still brand new at the time.
@@spacecitysprockets ah - 4 track compact cassette certainly does drift. And tape stretch, as I’m sure you know, is something of an occupational hazard. I remember we were all advised to not use BASF because apparently they stretched more. However, those cassettes are only running at 1.875 inches per second which is 16x slower than an typical multitrack. It’s those that I’m saying don’t really drift, assuming it’s properly maintained and setup. Certainly the SMPTE codes were pretty stable at 30ips. Very interesting comment though.
@@DistortThePreamp Thanks for answering my comments! I do remember in video editing suites during the 80s, we could only afford 1 generation on 1" tapes, so it was very touch and go to do layering (overdubs).
Great identification of the situation and concepts to visualize what hardware timing gives us - but there are plenty of tools in a DAW to deal with this issue - moving events in any direction you want. Of courre that could be tedious, but there are simple ways to adding timing randomness... As someone who has been sequencing since the days of the Atari ST - In general with MIDI sequencing and computers, the PPQ resolution (or setting if available) also determines where events happen in terms of MIDI timing resolution.
This is why I always prefer to record all my parts with midi into Ableton or any daw and DIAL in the amount of quantization for each part. It is almost an art of its own but it is very possible and rewarding with modern daw. A percentage of x% difference quantization can make or break a part based off a really good human recorded part. And no recorded part is ever the same so the quantizing is always different.
Love it! I recently decided to become a music producer and a DJ and among the first thing I acquired besides Pioneer FLX10 controller was a Roland TR-8S. I knew I was going to use Ableton for production and live performances, but my gut feeling told me I needed hardware instruments. My latest purchase was a Casio Privia S3000 with 88 fully weighted keys. Words cannot express how much I love that thing! A full piano keyboard lets me explore the full spectrum of harmonies as I try to find happy accidents while jamming along to my looping track.
Oh brilliant! I think the TR-8S is a great choice. With drum machines I’m a huge fan of trying to ‘perform’ with them - using mutes, solos, filters, etc. And the TR-8S is really good at that. My top tip would probably be to go easy on the fx. They sound like ear candy in headphones but live I would use delay as an occasional effect and reverb pretty much never. Obviously the filter should be used all the time though ;) And the S3000 is great. I have a proper 88 key piano but it’s not rigged up in my studio because I don’t have space for it. And I really miss it. If you’re a ‘player’ you really need those extra octaves. Let me know how you get on!
@@DistortThePreamp you know what else TR-8S is great at? Being an Ableton controller. You basically get up to 11 channels with dedicated volume fader, filter knob, reverb and delay! But unfortunately I have mot found a way to launch abd stop clips with it, so I am getting a launchpad mini next.
@@riccello oh that’s interesting! I used to use the Akai APC40 before I went dawless. It was really fantastic because not only could you launch clips but it had tight integration with things like effects racks. So from memory you could pre-map effect chains to racks and macros and then control those very easily. The rotary controllers were also very well done - infinite turning with leds to show values etc. It also had the A/B function implemented which not many people use in Ableton but which I used to use quite a lot.
@@DistortThePreamp btw, I was so inspired by your video, that I researched a way to simulate random phase drift in Ableton. Basically, you can do two things. Fiest, you can Add a delay to each channel and set a range from 1 to 20ms. Then you can automate it using LFO. You can also randomize LFO rate. Second thing you can automate the same way is phase shift between left and right channel.
Ah - randomising the LFO is a great idea! 20ms sounds a bit much though. I would set the max to maybe 12ms. And maybe have it changing value every 16th note. And obviously apply it separately to every drum. That’s a really interesting solution.
An additional tweak would be this: suppose you’ve set the max drift to 12ms, then you could also set a negative delay of the whole track to 6ms. So that the note was drifting around the grid line.
And I don’t reckon you’re gonna have to do anything with the phase - although it would be interesting. But the mere fact that the instruments are changing their alignment slightly every time is gonna make the whole thing sound different to a DAW. As I think I said in the video, it wasn’t so much that I could hear the sound, it was much more that I could hear the contrasting sterility of the strict grid.
It's not too dissimilar to what the Beatles did with their ADT (artificial Double Tracking) where they'd take the same vocals and bounce it to another track, but with a tiny delay to thicken things. Sure, in those cases the delay is enough for you to notice but it's the same sort of principle.
I'm not sure how or why The Pet Shop Boys did this in this case - it might be part of their MDI chain causing delays, or it might be how they edited takes together or something (I've also had hold of some of their multitracks in the past). Either way, serendipity and all that causes great things.
There's always that saying of make space in your music to give it space to breathe. So it's not only the case with long periods where other instruments can come in, but also the more subconscious stuff like this where a tiny delay that we can't instantly perceive can change the sound signature enough, much in the same way as playing with a phaser is tiny in amounts but you can quickly hear the difference.
Hahaha! I see what you did there ;) I couldn’t possibly confirm or deny what multitrack I had up. But really there were and are soooo many things that cause drift in the old analog workflow. Even CV/gate which everybody seems to believe is “rock solid” isn’t rock solid at all - of course the triggers transmit at the speed of light, but there a small computer spitting out the trigger timings and, well, turns out it’s a bit wonky. You’re absolutely right about phasing. I still maintain that I can hear the difference in the way a snare sounds against a kick because of alignment changes waaaaay before I can hear the actual changes in timing.
I didn't want to say it, but glad you noticed :)
Yup, I'm someone that for whatever reason dones't like using DAWs anyway. I trained at Salford Uni back in the early 1990s in audio engineering, so this was all original MIDI and Atari STs (if you were lucky) or analogue tape. I think I just prefer being more hands on.
But yeah, there's always lag on such things, especially when you extend the chain more and more. I do find it laughable that people call CV/Gate rock solid. They clearly weren't alive back in the day and had to deal with timing issues :)
By the way I love what you do - I appreciate talking about things that most people just well, don't.
Cool video! A few month ago I stumbled across the same problem and was astonished to see this. But I was editing the drums in a midi-track in the box (quantised it), played out to my RX11 and recorded it back via the seperate outs of the machine into the computer. Same effect. So my guess, it's somehow the midi-protocol that causes this. Or it is the "Midi-Synconisation-Protocol" that I've set to 50% default in the Computer . . . however, I know now about this issue and use it. Much better results. Thank you for your work. Greetings from Dresden.
@@le49exileaudioproduktion59 hey! So I thought it was the midi protocol too until I tried it with no midi at all. In other words just having my R8 spit out the drums when I pressed play. But whatever the reason, it sounds great so I’m not complaining!
@@DistortThePreamp interesting . . . hmmm. 😏
Fascinating video. I think I have noticed these subtle changes in timing, but never put my finger on it. "Temporal phase drift", that's a very clever term. I follow a producer called Mirko Hirsch who does a fantastic job with some of his songs at remixing songs into 80's style.
Fascinating theory! Thank you for your great analysis! I think it makes total sense, and it's something I've considered also. what I don't really know yet, is a good way to get this done in my DAW. Any advice would highly appreciated!
Ah yes, there are lots of good ways. If your DAW has a bumanise function then that would work. For example, in Ableton you can use a Max4Live device called Humanize, stick it on every midi channel, and make sure it’s calibrated to move the midi notes within a +/- 7ms range (which might take some testing). Logic has a similar function built it. There are other methods, but I think those are the easiest! Because of my own personal workflow I’m a huge fan of actually rendering out the audio before mixing so there’s no longer any midi in the session, and in this case you would be rendering out humanised midi. This makes sense to me because it would mean you were taking a snapshot of the randomised timing drifts rather than having them be different with every playback. But that’s just me ;)
@@DistortThePreamp Thank you! I'm very inspired to set this up in Cubase.
@@Tt-yr9ri Oh - absolutely let me know how you get on! Again, I strongly encourage you once you’ve found the right sound to ‘perform’ it in to Cubase as audio. In other words set the sequence running but if possible make some parameter tweaks while it’s playing and capture audio not just an automation line. Obviously preserve the midi, but then try to work with the audio. IMO committing in this way always helps a production ;)
@@DistortThePreamp It actually worked :D. There's something called Logical Editor i Cubase; it has presets to offset MIDI notes. Now it sounds a bit more as if my vst synths are performed by a band :D.
Oh brilliant! My top tip is not to overdo it. A very small amount goes a long way. I don’t know if you could measure the effect, but my recommendation is that every instrument, particularly drum instrument, moves a maximum of 7ms in each direction, and of course they have to move independently. That’s not to say that larger drifts won’t have an interesting effect too! But to just get the thing where every combination of kick and snare is technically different, you only need +/-7ms. Very exciting that you’re trying this though!
6:33 Magic microphone trick. Love it!
Ah yes. I had a wizard cast a spell which means that the mic no longer needs to be pointing at my mouth to pick up perfect audio. Very useful for UA-cam videos.
This was fascinating... And the stethoscope makes you look like a musical eccentric or batshit bonkers. 👍
Hahaha! Remember, the stethoscope doesn’t have to be vintage, but it does have to be medical grade 😂
Great explanation!
I don't have the technical knowledge to understand the time deviations in drum machines, but I know that nothing in nature is perfect. Our primitive instinct picks up on these imperfections, which gives us a sense of "trust," while things that are too perfect can raise suspicion and cause rejection. That’s why those subtle imperfections are more pleasant to us; it’s a clear example of the uncanny valley effect.
Cheers!
That’s exactly right. My theory is that even though we can’t hear the time differences, we can sense the sound differences because we have evolved to sense minuscule sonic changes arising from alignment since that’s how we detected sounds coming from behind us (and avoided getting eaten). FWIW all the producers I showed this too (some of whom have won Grammys and been making records for 40 years) were incredibly sceptical until they tried it in their own studios with their own drum machines.
I've never liked the sound of drums being exactly on the drum grid, so I have always played my drums by hand and then used midi quantize around 60-80% (until it feels right). I'd imagine the resulting time variations are roughly similar the 3-7 ms variations produced by hardware sequencers, albeit perhaps less random.
Yes I think that’s right. Even a really tight human drummer is a few ms out. I mean, let’s start with the fact that the snare is closer to their ears than the kick, and it takes sound about 1ms to travel a foot!
I give your videos the thumbs up even before they start. It's so true that computers and vst will never emulate hardware.
Oh that’s amazing! Thank you! And of course you’re completely right about hardware emulation. So many tiny things.
Another excellent contribution
Thank you! This one stuttered a bit, basically because (I think) I had the wrong thumbnail so a lot of people missed it. So thank you so much for watching!
It doesn’t matter if it’s a software or hardware sequencer the delays are inherent in the serial nature of MIDI.
Absolutely true. But, here’s the kicker, there was no midi. I tested all sorts of configurations, but the one that really confused me was when I just hit play on the drum machine and still got drift. Not only that, but some arrive early. It’s *definitely* drift at the drum machine/sequencer level.
@@DistortThePreamp I produced a lot of music in the late 80s, early 90s and used more equipment than I can remember. One of my drum machines was the R8 and I can confirm that it reacted differently when using its own sequencer compared to being triggered. Although both had issues. Everything has a delay/latency, keyboards have latency, sequencers etc. There’s even latency on different I/O of the same type. The period of time you are talking about “80s drums”, presumably the earlier half, would mostly have been CV/gate triggering for sequencing, which had its own latency issues, although, similar to MIDI in it being serial. Then you also have to consider the type of tape synch being used to synchronise CV/gate sequencers to tape, which was likely FSK, which would have had its own issues due to a relatively low resolution pulse. Which would have also suffered issues with its square (pulse) wave being rounded due to the many issues with tape. Even once we started using SMPTE to MTC, the issues of tape synch were still part of the overall latency equation. In short, the issue you are talking about is not unique to any time period of music creation with sequencers. These offsets you talk about where also part of the early 90s dance sound. Most of us doing it seriously would have been using something like a Midex on an ST to get multiple parallel outputs, which was only used to mitigate the timing issues as much as possible but there were still there. Then compound that with SysEX and CC data and the issues are still very much relevant today. Like I said, the timing errors you are talking about are not unique to the period of music you are talking about. Any “sound” differences from one decade to another are going to be primarily due to the recording and mixing technology and less about the actual sound making machines and their inherent issues. Or put it another way. If The Human League had have recorded and mixed Being Boiled or Empire State Human using the exact same synths but recorded and mixed them 10 years later, they would have sounded very different. Cheers, Dave.
Great video! Can I ask a question please? I have some synths and drum machines from the 80's. I'm a huge RX11 fan! I like to use either Cubase or Cubasis on my iPad to sequence them because I have noticed slight timing discrepancies that I like very much. You think this would be exaggerated by using the drum machines sequencer?
@@JamieMallender Interesting question. In my experience it’s different but not necessarily exaggerated. And to be honest all you really need is enough of a timing discrepancy that every combination of hits sounds ever so slightly different to the brain which, again in my experience, seems to happen before you actually notice definite timing errors. I suppose that’s a long way of saying I don’t think it would be exaggerated. But, even if it were exaggerated I don’t think it would be better. One thing you should absolutely experiment with though is tracking in the audio from the drum machine and then fitting the grid to whatever audio was captured (usually the kick drum) so then you have a daw grid but it’s following the crazy timing. I do this by creating a tempo map. This makes for an extremely flexible workflow and basically means that you should even be able to overdub drum patterns having ‘taken’ the drums. What drum machines and synths do you have btw?
@@DistortThePreamp That's very interesting. I shall definitely experiment with that, thank you. I have a TX7, Poly 800, FB-01, RX17, RX11, DR-550, a soundmaster SR88 and from the early 90's a DR-5 and a QY22. I've used some of these things in my videos. I like composing with a combination of hardware and software.
That’s a very interesting gear list. I wonder if you might like a rompler like the Proteus 1 to go with your FM and analogs? I don’t think romplers sound good enough to build a whole song around, but they can be fantastic for layers or supporting sounds. You basically use presets (with a bit of tweaking) and the obvious ones are the Emu Proteus, the Korg M1, or the Roland JV 1080 (or the 2080). It sort of depends on what kind of music you make, but I’ve got the first two and really wish I had a JV. As I said, I would never rely on them for a whole track, but if I want to reach for strings, brass, vibes, etc they are a great thing to try. Just a thought…
@@DistortThePreamp am I right in saying the Proteus is a rack mount? That could work for me. I don’t have room for any full size synths. My studio is full to bursting mostly with guitars and basses. But yeah, that sounds like it might be a solid purchase, thank you. ✌️✌️
Pet shop Boys?
Brilliant as always. Great to see your channel growing
9:33 Actually, that's true, but there is also a simple solution for the computer to bring more life into the tracks (especially to simulate this behavior). In Ableton, this is achieved by combining track delay and the echo effect with an LFO tool. Since the track delay cannot be automated, we use the delay instead. You insert a simple delay effect and switch it from beats to time. Then, set the effect to 100% wet and drag the time parameter all the way down to 1ms. Switch the mode to fade and Feedback to 0.
Now, open an LFO and route it to the timing parameter, adjusting it to, for example, 8% and 10%. This will gently modulate the timing parameter, allowing us to mimic subtle timing inaccuracies. To make the modulation more random, set the waveform to Sample and Hold in the LFO tool (Random)-and voilà! It shift´s the track very sattle within 1ms etc. You can decide and play around with the LFOs speed settings....
This can also work in the negative direction. For this, drag the track delay parameter to for example -10ms and set the delay to 10ms. At this the tracks will be synced and play around with the LFO there :)
Thank you for your video and if you have questions about it, you can simply ask. Best regards.
Thanks for this. Yea you’re absolutely right. The method I thought for Ableton is to use a neutral groove with a timing randomisation on the 16tg notes. But actually as a few people have noted the M4L humanise device might be even better. I’m not sure about an LFO because it would cycle (although in context that might be fine) although, as I’m writing this, doesn’t the Ableton LFO tool have some kind of S&H random setting? All good. And it goes without saying I was being a big flippant when I said computers were dead. What I really meant was that if all the transients are on the grid then that’s “dead”. But I know that even that depends on genre - deadmau5 is pretty grid based (although he advances his claps and snares) and his music doesn’t sound lifeless at all! Really appreciate you watching and taking the time to write this comment :)
@@DistortThePreamp Thanks for your work! Ableton's LFO has a random function, which can be found above the rate control where it normally says 'Sine.' Click on that and set it to 'Random,' which simulates Sample & Hold. I watched the video and thougt, this could be a possible solution and build it - it worked.
Oh fantastic!! Did you have to do much screwing around to get the delay small enough?
@@DistortThePreamp No, not at all. You have to switch it to time and turn the dial to full left, it will be at 1ms and from there you can adjust it to the right. If you need, you can hold shift and make very sattle changes with the mouse.
One word: humanize. This was a feature of DAWs for like 30 years
And it works very very poorly. And actually sounds very much like what it is. Randomness.
@@vladv5126 well that's exactly what this video describes though, random offsets from the grid. You just have to set it so that the offsets are no more than a couple of milliseconds.
Probably better to record the audio of the "tight" drummachine and then use effects to shift the microtiming, then to do it in MIDI, since MIDI notes can only be moved in form of ticks which isn't as fine as the audio grid or better: delay times.
ANd you never get the original. Any ideas of songs made with dawn that got the soul of 80' s hit ?
@@mikosoft it's not quite the same. For some reason there's a pattern to the randomness of these old machines because they're actually trying to be consistent. The humanize feature in daws does the randomization all too well. It's actually best to just play the parts by hand with a controller and not quantise.
How to introduce temporal phase drift in Ableton? I have no hardware
I’ve been thinking about this a lot. Here’s what I would do:
1. Make sure your drums (and other midi instruemnts) are all on separate tracks.
2. Load up a neutral 16th note groove from the groove pool that basically does nothing
3. Add some randomisation to the timing. You might have to experiment to make sure that the randomisation sticks to within about 6ms either way.
4. Apply this to each of the drums and instrument tracks.
I haven’t tried this, but it should work. The important thing is that the instruments and drums drift AGAINST each other
@@DistortThePreamp thank you I will try this! A producer named Noisia who produces some of the best sounding drum and bass tracks stresses his preference for making things sound organic and in motion so I am excited to implement your info as I struggle making my sound feel alive (let alone sound good!). I love the 80’s sound, my doodles always gravitate towards that style, thank you for all the videos! Your Duran Duran arpeggio video was amazing.
Oh thank you! Yes, I think people focus on timing to make things sound organic, but overlook the microscopic sound changes (that are sometimes a byproduct also of timing). I’ve had so many people asking me the “how do I do this in Ableton?” question that I might take a deep breath and figure it out.
I think it will work, but if it doesn’t then another possibility (maybe a last resort) would be for me to create and give away a whole load of 16th note grooves (taken from my actual setup) for people to use, but with the idea that no two instruments use the same groove. Hopefully it won’t come to that because that sounds like a rather over-elaborate solution.
Do let me know how you get on though.
Nothing wrong with using a DAW, but instead of quantizing the recording you'd have to humanize it instead.
And even if you'd quantize the MIDI-track for drums, the moment when you send that MIDI-signal to your Behringer RD-8 (TD-808-clone) and record the audio output, you will have these slight timing imperfections.
Oh, and also think of the slight speed variation you'll have when you record with an analog multitrack tape recorder.
You’re of course 100% right about the DAW - all possible using humanise. But in fact there was no midi with the drum machine - it was just running free!
@@DistortThePreamp yeah, it just used an internal clock running at x bpm. And the rest (bass, guitar, keys) had to follow that pattern by ear, leading to more slight timing imperfections. This causes a natural groove and makes music come alive.
Also, weren't they all synched to an SMPTE track on the tape machine? Wouldn't that do more to account for drift than anything to do with the clock in the drum machine?
@@replaceablehead Yes, that might account for some drift in overall time, but not changes in alignment between hits.
Whats the nature of the drift in the music you analyzed? Is it that each instrument is offset by a particular amount, or are they jittering so at any beat different instruments are triggering at randomly different times, or do rhe different instruments drift in and out of sync over the song like there's a slow LFO?
I want to know more of your science!
Great question. They’re jittering so that the offset varies each time, including the direction. Rarely more than 7ms in either direction, but also never ever aligned. The fact that they’re never aligned leads me to believe that drum machines have some sort of serial architecture and there’s a minimum spacing of 1ms (similar to midi). It’s also random rather than an oscillator. For the experiments I ran the drum machine(s) on their own, and also as sound sources sequenced by external sequencers. I didn’t go into the detail in video, but tye timing got worse when external sequencers were throwing more notes at the machine. That seemed to matter less when the machine was internally programmed. All fascinating. I freaked out when I discovered all of this, but now I’ve decided that it’s part of my production workflow.
Interesting that external sequencing causes bigger jitter, and it's (apparently) proportional to the amount of notes. With some back of the envelope maths, MIDI is 32.1kbps, it's 10-bits per byte (with start & stop bits), and note on & note off are each 3 bytes, so sending or receiving a note or off will take almost exactly 1ms. Assuming everything is perfect and instant (when is it ever?), if your sequencer is sending a note-on message on three different channels on the same bus, the first and third note-on messages will be received 2ms apart. If in your sequence there are preceding notes ending right as the new ones start then for each note-on the sequence will first send a note-off and that 2ms doubles to 4ms. 4ms is equivalent to 250hz, so you're definitely in the realm of phasing. Very cool!
Yeah, this is exactly right. My suspicion is that internally the drum machines’ own sequences are implemented in a way that is either straightforward midi or similar to midi. Even the modern Behringer RD9 is incapable of spitting out hits that land together, and that’s when you press play on an internal sequence with no midi. Everything is always at least 1ms apart which I think is telling.
And when I’ve got a huge sequence running for a song I might easily have 10-20 things happening on the same beat, albeit on different midi channels. I like the timing drift but that can get a bit too sloppy, so I often turn off the extraneous stuff when it comes time to track everything in, just to keep it a little tighter.
@@DistortThePreamp Wait, you're getting jitter? How are you measuring? I recorded my RX5 and then got reaper to slice the transients at a certain threshold and then I measured each slice. This method had the RX5 measuring accurate to with 50 samples at 44.1khz. Have you read the Roger Linn interview?
@@replaceablehead Yes I have read it. These were measured in Pro Tools, at 48kHz. Not only were there timing shifts but, much more importantly, the alignment between drum instruments changes. I’m not discounting Roger as a technician, but it’s worth point out that he has never acknowledged the effect that the inaccurate sampling in the Linns had on the effective timing. Because the samples weren’t sliced at the transients, the gap at the front varied as you detuned each drum. So literally every combination of drum tunings would have a slightly different feel. This was very pronounced indeed on the rimshot when it was severely detuned by, say, Prince.
Are hardware sequencers, like the Yamaha Motif xs also capable of creating that unique feel simply because they’re external from the computer? I often find that my inspiration and results are entirely different when I record from a hardware sequencer into my computer. Something about it not aligning perfectly with the grid brings a kind of magic I really enjoy. Is that the intention behind using hardware?
Interesting question. I haven’t tried the Motif but it very likely does the same thing. I suppose if it’s only sequencing internal sounds then it might not, but certainly if it’s sequencing external sounds it’s very likely to have the same “issue”. In terms of your question “Is that the intention…” I think that for me it’s one of the reasons I like using hardware so much. My original reason was all about the workflow - the physically of pressing buttons and turning knobs is very different to sitting in front of a computer monitor, and in my experience producers much better results, much faster. But the timing revelation opened up a whole new set of reasons to prefer the workflow. I can’t say it’s definitively better as there’s lots of ‘grid music’ that I love, but I think it’s definitely ‘better for me’, if that makes sense ;)
The key to sounding like the 80s is writing actual songs.
My best tracks are made by a mix of machine sequencing, hand playing and some sound processing synced up to one of the machines (that I somehow have to hand start). .
That sounds very similar to my workflow!
Just found this. Love it. You have a new subscriber. Now, gotto buy a stetoscope. Have a nice weekend! Best , old-school electronic musician ai accident
Oh fantastic! Make sure you get a vintage stethoscope - the modern ones don’t work ;) I’ve been literally editing all day (12 hours non stop) so this is a nice message to come out of the edit to 🙏
Thank's for that Video. You bring it on the Point. In my first experimental Recordings from 30 years ago I useed drum machines and sequenzed Parts triggered by the machines and it sound always intersting and human cause there is no masterclock by a computer. Today my Korg DDD 1 is the best tool to programming beats cause it has it's own heart and the sounds are as they are. 80ssssss. Love, peace and harmony Bernward
Oh excellent! Yes - all about there being no masterclock! So glad you liked the video :)
Old drum box had much more jittering like Roland TR 808 / 909; Linn LM1 / Linndrum. Modern clones completly lost this as they are clocked with modern processors, LM1 sequencer was driven by a VCO .
So this is the funny thing - I haven’t tested the modern Roland drum machine but my Behringer RD-9 absolutely has jitter! And from what I can tell so does my Squarp Pyramid. The latter is extremely hard to prove because there are so many other factors but I did some midi-free tests using the CV/gate and there was still drift. It’s all really astonishing. My best guess is that although they all have quartz clocks which in theory are extremely accurate, there’s some cheap chip being used to handle the timing. In truth quartz clocks were also used in the old units back in the 80s - the tech has always been there, it’s just that it was compromised by design choices (is my educated guess). For example, the Synclavier was only ever accurate to 10ms which is quite shocking since they marketed it as totally accurate. Etc etc etc. And of course the final irony is that in most cases the drift sounds good and makes the sound combinations vary.
Great info! I did a Podcast Ep a few years ago about playing on the grid. "Groove" can be a complex topic. I'll add this: as I Zoom in on the GB/Logic Drummer, the hits have "temporal phase drift". The Humanize dial in this case is about 5%. With it off, the beats are lined up with each other but slightly ahead of the Grid. The dial for me is a bit unpredictable but it's something to check out. Enjoying your channel.👊🥂
Oh that’s extremely interesting! And they’re not aligned, right? So the drift is on a per instrument basis? I think 5% sounds like a lot. I don’t know how the percentage is being calculated but you really only need a few ms of drift to get the effect. So glad you like the channel. Really appreciate the comment :)
@@DistortThePreamp That's right- Not Aligned. The Humanize dial does not give you a number (I wish it did but I'm guesstimating, based on a full turn) so I sometimes zoom to a fixed amount and eyeball the instrument displacement.
That’s a totally solid technique. There’s definitely a point when the brain can’t quite hear the timing differences but it can sense that the sounds are changing and the music comes ‘alive’. Very important IMO.
Small counter argument: sequencing hardware _from_ a computer will incur MIDI jitter. This might be a problem if you want really tight hi-hats but otherwise I say embrace it.
I remember reading an interview with BT (the dance music producer) maybe 15 years ago where he pointed out the same thing, but the opposite way round. He was a bit obsessive about making stuff perfectly sample accurate, laying it out as audio on the Logic timeline, because he'd noticed MIDI timing was all over the place and sounded dated.
Yes, this is what I used to do. We all used to talk about midi being ‘sloppy’ and we’re obsessed with being ‘sample accurate’. I’ve now gone the other way 😂
What is interesting about this is that he is right have you ever heard a drum machine miss a entire note and leave you with a very small piece of static fuzz that is barely perceptible. It won’t happen very often but when it does it will shock you… maybe
Yes! My R8 gets confused sometimes and drops snares. It would be hilarious apart from when I’m tracking and don’t notice. Then I have to give it a strong telling off.
I noticed timing variations in the 90s with MIDI. The track arrangement in the sequencer was such that first drums, then bass, then solo, then pads (these were delayed again) came one after the other, to get the groove somewhat tight. This led to a certain 90s dance vibe. I don't know the situation in the 80s. But I guess that in addition to MIDI, there were also sync fluctuations and fluctuations in the devices, which explains the pre- and post-delay.
That’s what I would have thought, but actually it’s baked into the drum machines. Those measurements were just from me programming a pattern and hitting play. No midi involved. Totally erratic.
@@DistortThePreamp Makes sense. Something that wouldn't have been on my radar at first. At second glance, of course, the processing speed of the commonly used 8-bit CPUs was also limited in order to fire samples perfectly on a grid.
Yes. It's particularly ironic with the Roland R8 drum machine -- that was promoted as having extremely high resolution. Instead of 32 notes per bar you could do 960 so you could, effectively, programme actual triplets and very detailed rhythms. Who knew that the actual timing error margin was way over this resolution. Not that it's not legit -- it's all legit. But, honestly, nobody knew, because nobody checked.
@@DistortThePreamp it would be really interesting if Roland knew that their older machines had timing inconsistencies, found the R8 to be too precise and re-introduced the inconsistencies so as not to lose the magic.
I've been aware of this for a while, I play everything by hand to combat it
Very, very good idea
So what about the Fairlight CMI? It was basically a computer and was used quite a lot for making beats in the 80's (e.g. Opportunities by the Pet Shop Boys). It doesn't get more eighties than that...
Yes. Although there was still a surprising amount of analog midi gear involved in those records. Same with the Synclavier which had its own timing problems as well. It was marked as being entirely accurate, and it turned out what they meant was accurate to within 10ms.
Brillaint. Another perfect tip. Cheers!
Thank you! 🙏
Great video ! Very interesting informative ! Thank you ! :-)
Oh awesome! Really appreciate you watching and taking that time to comment :)
Surely those timing errors are trivial to program into a modern DAW to recreate that 80s sound? Also, some audio examples to listen to would have been nice. I've seen your reply to someone else saying you're thinking of remaking this video with some samples, so looking forward to that.
Make some manual adjustments that take a few drum hits slightly off the grid and you are golden.
I can recommend the Akai MPC 5000 as your main hardware sequencer. It has a very sloppy timing and sounds absolutely gorgeous on it's own.
I have a roland sh-01a (the modern recreation) which I send midi to from my DAW and then record the audio. I've noticed that the timings of percussive sounds vary quite considreably, by several milliseconds, and perhaps up to about 10. It's possible to see this variation in timing against the grid even when you have say 4 bars of audio across the screen. At first I was a bit perplexed by this and thought it must be coming from the midi as the synth is basically a VST in a box, or so I thought, but maybe this is Roland's way of making it sound a bit more analogue. Sometimes it does seem to add a certain something, but on other occasions it's downright annoying and I have to go in and manually adjust the timing.
Another great vid, keep up the good work!
I love your channel, always passionating ! You are the eighties ! 😂 (and more...)
Hahaha! I think Espen has a copyright claim on that 😂😂😂 Thank you though…
Well ai got it right. I was overlaying drum loops I generated with udio ai in the daw (mostly 70's and 80's dicso generated stuff) and noticed that my kicks with the exact bpm as the udio tracks don't align properly with the udio stems. Hense if your cpu and daw is crap you could also put vst effects that doesn't do anything on separate drum samples mixer tracks and they will introduce latency and will cause phase issues too.
It's funny how some people obsess these days about timing just like pitch perfection. I never thought there was anything wrong when I was slaving my Atari to my Tascam 244. I learnt back then to avoid quantization.
However, the problem I had was eagerness to fix flubs in Cubase instead of practicing till I got things right - a very bad habit that I carried forward to digital audio multitracking that not only stunted my growth as a musician but resulted in a vicious cycle of increasingly lacklustre performances.
Yes, that’s exactly right. Very important not to spend hours fixing and editing and you can re-take in about 5-20 minutes.
My Roland MC50mkII still here
Oh wow! That’s the sort of sequencing cousin of my Roland R8 that was released around the same time. I don’t know it very well - I’m familiar with sequencing drums on the R8 which is a dark art, and I have a Roland MSQ-700 sequencer which is another dark art, but I bet it’s really good once you get to grips with it. Is that what you’ve found? You still use it!
Truth…? Let’s go anecdotal:
Gerhards Lengeling’s precursors to Logic (e.g. Creator/Notator) ran on C-64 and Atari ST. Sync to tape (the only option, not phase perfect) was bound to recorded linear time code (e.g. SBX-80).
A Linn 9000 synced to tape from a recorded square wave, counting pulses to BPM, maybe at 96 pulses per division. The Synclavier would use an even more expensive sync-unit with quartz PLL - but still the same tech as the above.
All of this happened on complex “CPUs” less powerful than 4$ microcontrollers - Then, an 8-bit serial protocol (MIDI) is sequenced onto this heap of uncertainties…?
That’s how we did this, at Polar in Stockholm or at Advision in London. Some lucky few, in their own studios. There’s no magic, just our silly grinding to get around the limitations of the best tools available for labels’ or publishers’ cash.
As fun and frustrating as recording music has ever been. Learn your tools (that’s what they are). The magic is within yourselves, as musicians interacting - not inside retro boxes of unobtainium, barely able to crunch serial data… 😅👍
i like to compose midi on the computer sequencer then record the midi to the hardware sequencer for recording or live performance, the clock on these machines have a vibe as you say, i tried working in the box for years i could never get that timing/feel, every hardware sequencer has it's own feel going on.
Oh that’s a very good plan. Interesting workflow. Will sound very different indeed.
@@DistortThePreamp sounds tedious but i found it the best workflow to get the results i want, none of my harware sequencers have screens so edits at the note level can be a pain, although i do like using my step sequencer to jot out ideas, ti like the computer sequencer for are editing (prog changes, cc data, sysex) and saving work, then i record that to the machines, works a treat. : )
I've messed with using random midi effects in Cubase to do this. Bit of timing and velocity variance on a sequenced pattern goes a long way. I mostly use it for arpeggiated piano though, like others have said it would never occur to me to do it for an 80's song.
Temporal Phase Drift sounds like something Dr Who would know all about. :)
Great video love the detailed explanation and the mysterious questions haha. I use Fl studio and I know on each drum sample has an option to let you offset the sample. Would that be a good way to emulate the slight time differences you spoke on? I
Not quite as it would mean the offset never varied. But I’m sure you could rig something up with a slightly randomised groove… :)
”Humanize” quantization found in many software
@@DistortThePreampnot even if you did to each individual drum sample? Thanks for the response love your channel 💜